Skip to main content
main-content

Über dieses Buch

This two-volume set LNCS 9771 and LNCS 9772 constitutes - in conjunction with the volume LNAI 9773 - the refereed proceedings of the 12th International Conference on Intelligent Computing, ICIC 2016, held in Lanzhou, China, in August 2016.

The 221 full papers and 15 short papers of the three proceedings volumes were carefully reviewed and selected from 639 submissions. The papers are organized in topical sections such as signal processing and image processing; information security, knowledge discovery, and data mining; systems biology and intelligent computing in computational biology; intelligent computing in scheduling; information security; advances in swarm intelligence: algorithms and applications; machine learning and data analysis for medical and engineering applications; evolutionary computation and learning; independent component analysis; compressed sensing, sparse coding; social computing; neural networks; nature inspired computing and optimization; genetic algorithms; signal processing; pattern recognition; biometrics recognition; image processing; information security; virtual reality and human-computer interaction; healthcare informatics theory and methods; artificial bee colony algorithms; differential evolution; memetic algorithms; swarm intelligence and optimization; soft computing; protein structure and function prediction; advances in swarm intelligence: algorithms and applications; optimization, neural network, and signal processing; biomedical informatics and image processing; machine learning; knowledge discovery and natural language processing; nature inspired computing and optimization; intelligent control and automation; intelligent data analysis and prediction; computer vision; knowledge representation and expert system; bioinformatics.

Inhaltsverzeichnis

Frontmatter

Evolutionary Computation and Learning

Frontmatter

A Hybrid Scatter Search Algorithm to Solve the Capacitated Arc Routing Problem with Refill Points

This paper presents a hybrid scatter search algorithm to solve the capacitated arc routing problem with refill points (CARP-RP). The vehicle servicing arcs must be refilled on the spot by using a second vehicle. This problem is addressed in real-world applications in many services systems. The problem consists on simultaneously determining the vehicles routes that minimize the total cost. In the literature is proposed an integer linear programming model to solve the problem. We propose a hybrid algorithm based on Scatter Search, Simulated Annealing and Iterated Local Search. Our method is tested with instances from the literature. We found best results in the objective function for the majority instances.

Eduyn Ramiro López-Santana, Germán Andrés Méndez-Giraldo, Carlos Alberto Franco-Franco

A Novel Fitness Function Based on Decomposition for Multi-objective Optimization Problems

Research on multi-objective optimization problems (MOPs) becomes one of the hottest topics of intelligent computation. The diversity of obtained solutions is of great importance for multi-objective evolutionary algorithms. To this end, in this paper, a novel fitness function based on decomposition is proposed to help solutions converge toward to the Pareto optimal solutions and maintain the diversity of solutions. First, the objective space is decomposed in a set of sub-regions based on a set of direction vectors and obtained solutions are classified. Then, for an obtained solution, the size of the class which contains the solution and an aggregation function value of the solution are used to calculate the fitness value of the solution. Aggregation function which decides whether the target space is divided evenly plays a very important role in the fitness function. A hyperellipsoidal function is designed for any-objective problems. The proposed algorithm has been compared with NSGAII and MOEA/D on various continuous test problems. Experimental results show that the proposed algorithm can find more accurate Pareto front with better diversity in most problems, and the hyperellipsoidal function works better than the weighted Tchebycheff.

Cai Dai, Xiujuan Lei, Xiaofang Guo

MREP: Multi-Reference Expression Programming

MEP is a variant of genetic program applied to solve the symbol regression and classification problems. It can encode multiple solutions of a problem in a single chromosome. However, when the ratio of genes reuse is low, it may not get a high accuracy result within limited iterations and may fall into the trap of local optimum. Therefore, we proposed a novel genetic evolutionary algorithm named MREP (multi-reference expression programming). The MREP chromosome is encoded in a two-dimensional structure and each gene in one chromosome can refer other sub-layer’s gene randomly. The main contribution can be described as follows: Firstly, a novel chromosome encoding scheme is proposed based on a two-dimensional structure. Secondly, two different cross-layer reference strategies are designed to enhance the code reuse of genes located at different layers in one chromosome. Two groups experiments were conducted on eight symbol regression functions. The statistical results reveal that the MREP performs better than the compared algorithms and can solve the symbol regression functions problem efficiently.

Qingke Zhang, Xiangxu Meng, Bo Yang, Weiguo Liu

Independent Component Analysis

Frontmatter

Extraction of Independent Components from Sparse Mixture

In this paper we study extraction of independent components from the instantaneous sparse mixture with additive Gaussian noise. We model the problem as a dictionary-learning-like objective function which tries to discover independent atoms and corresponding sparse mixing matrix. The objective function involves fidelity term, L1 normalization term and Negentropy term which respectively limits noise, maximizes the sparseness of mixing matrix and non-Gaussianity of each atom. An alternative iteration algorithm is proposed to solve the optimization. According to our simulation, the proposed method outperforms FastICA and K-SVD.

Jian-Xun Mi, Cong Li, Chao Li

Compressed Sensing, Sparse Coding

Frontmatter

Leaf Clustering Based on Sparse Subspace Clustering

Leaf recognition is one of the important techniques for species automatic identification. Because of not needing prior knowledge, clustering based method is a good choice to accomplish this task. Moreover, the high dimensions of leaf feature are always a challenge for traditional clustering algorithm. While the Sparse Subspace Clustering (SSC) can overcome the defect of traditional method in dealing with the high dimensional data. In this paper we propose to use SSC for leaf clustering. The experiments are performed on the database of leaves with noise and no noise respectively, and compared with some conventional algorithm such as k-means, k-medoids, etc. The results show that the clustering effect of SSC is more accurate and robust than others.

Yun Ding, Qing Yan, Jing-Jing Zhang, Li-Na Xun, Chun-Hou Zheng

A Compressed Sensing Based Feature Extraction Method for Identifying Characteristic Genes

In current molecular biology, it becomes more and more important to identify characteristic genes closely correlated with a key biological process from gene expression data. In this paper, a novel compressed sensing (CS) based feature extraction method named CSGS is proposed to identify the characteristic genes. Considering the transposed gene expression matrix and class labels as sensing matrix and measurement vector, respectively, CS reconstruction is implemented by basis pursuit algorithm. Top ranking genes with high signal weights are retained as the characteristic genes. Experiments of CSGS are performed on leukemia data set and compared with other sparse methods. Results demonstrate that CSGS is effective in identifying characteristic genes, and is not sensitive to parameters. CSGS could offer a simple way for feature extraction and provide more clues for biologists.

Sheng-Jun Li, Junliang Shang, Jin-Xing Liu, Huiyu Li

Social Computing

Frontmatter

Enhancing Link Prediction Using Gradient Boosting Features

Link prediction is an important task in social network analysis. The task is to predict missing links in current networks or new links in future networks. The key challenge in link prediction is being lack of features when machine learning methods are applied. Most relevant studies solve the problem by using features derived from network topology. In our work, we propose a novel feature extraction method by employing Gradient Boosting Decision Tree (GBDT), which effectively derive attributes from initial feature set. For GBDT model, input features are transformed by means of boosted decision trees. The output of each individual tree is treated as a categorical input feature to a sparse linear classifier. Extensive experiments demonstrate that the proposed method outperforms a number of mainstream baselines when GBDT features considered. The proposed method is efficient to solve the feature shortage problem in the prediction of links.

Taisong Li, Jing Wang, Manshu Tu, Yan Zhang, Yonghong Yan

Neural Networks

Frontmatter

Supervised Learning Algorithm for Spiking Neurons Based on Nonlinear Inner Products of Spike Trains

Spiking neural networks are shown to be suitable tools for the processing of spatio-temporal information. However, due to their intricately discontinuous and implicit nonlinear mechanisms, the formulation of efficient supervised learning algorithms for spiking neural networks is difficult, which has become an important problem in the research area. This paper presents a new supervised, multi-spike learning algorithm for spiking neurons, which can implement the complex spatio-temporal pattern learning of spike trains. The proposed algorithm firstly defines nonlinear inner products operators to mathematically describe and manipulate spike trains, and then derive the learning rule from the common Widrow-Hoff rule with the nonlinear inner products of spike trains. The algorithm is successfully applied to learn sequences of spikes. The experimental results show that the proposed algorithm is effective for solving complex spatio-temporal pattern learning problems.

Xiangwen Wang, Xianghong Lin, Jichang Zhao, Huifang Ma

Behavior Prediction for Ochotona curzoniae Based on Wavelet Neural Network

Ochotona curzoniae is one of the main biological disasters in the Qinghai-Tibet plateau and adjacent areas in China. Video-based animal behavior analysis is a critical and fascinating problem for both biologists and computer vision scientists. The behavior prediction for Ochotona curzoniae is a basis of Ochotona curzoniae behavior analysis in video recordings. In this paper, a three-layer wavelet neural network is proposed for short-term Ochotona curzoniae behavior prediction. A commonly used Morlet wavelet has been chosen as the activation function for hidden-layer neurons in the feed-forward neural network. In order to demonstrate the effectiveness of the proposed approach, short-term prediction of Ochotona curzoniae behavior in the natural habitat environment is performed, and we analyze the influence on prediction accuracy at various numbers of input neurons. The forecasted results clearly show that wavelet neural network has good prediction properties for Ochotona curzoniae behavior prediction compared with BP neural network. The model can assist biologists and computer vision scientists to create an effective animal behavior analysis method. The principle of our Ochotona curzoniae behavior prediction used wavelet neural network is helpful to other animal behavior prediction and analysis in video recordings.

Haiyan Chen, Aihua Zhang, Shiya Hu

New Filter Design for Static Neural Networks with Mixed Time-Varying Delays

This paper focuses on designing a $$ H_{\infty } $$ filter for a class of static neural networks with mixed time-varying delays. Here the mixed time-varying delays contain both discrete and distributed time-varying delays. Based on a Lyapunov-Krasovskii functional combined with a zero equation, a suitable $$ H_{\infty } $$ filter is obtained for the static neural networks model. The filter can be solved by a linear matrix inequality (LMI). Two numerical examples are presented to validate the proposed method. In addition, the obtained filter can be applied to design the control systems with delays.

Guoquan Liu, Shumin Zhou, Xianxi Luo, Keyi Zhang

Nature Inspired Computing and Optimization

Frontmatter

SMOTE-DGC: An Imbalanced Learning Approach of Data Gravitation Based Classification

Imbalanced learning, an important learning technique to cope with learning cases of one class outnumbering another, has caught many interests in the research community. A newly developed physical-inspired classification method, i.e., the data gravitation-based classification (DGC) model, performs well in many general classification problems. However, like other general classifiers, the performance of DGC suffers for imbalanced tasks. Therefore, we develop a data level imbalanced learning DGC model namely SMOTE-DGC in this paper. An over sampling technique, Synthetic Minority Over-sampling Technique (SMOTE), is integrated with DGC model to improve the imbalanced learning performances. A total of 44 imbalanced classification data sets, several standard and imbalanced learning algorithms are used to evaluate the performance of the proposal. Experimental results suggest that the adapted DGC model is effective for imbalanced problems.

Lizhi Peng, Haibo Zhang, Bo Yang, Yuehui Chen, Xiaoqing Zhou

Genetic Algorithms

Frontmatter

Solving the Static Manycast RWA Problem in Optical Networks Using Evolutionary Programming

The Static RWA (Routing and Wavelength Assignment) problem in Optical Networks is a combinatorial optimization problem fit to iterative search methods. In this article we further investigate the static manycast RWA problem in optical networks and solve it using an evolutionary programming (EP) strategy such that the number of the manycast requests established for a given number of wavelengths is maximized. The proposed algorithm solves, approximately, the wavelength assignment problem while a backtracking approach is used to solve the routing issue. We present the details of our proposed algorithm and compare it to another metaheuristic named genetic algorithm (GA). EP shows a 24 % improvement over GA.

Amiyne Zakouni, Jiawei Luo, Fouad Kharroubi

Signal Processing

Frontmatter

A Control Strategy of Depressing the Voltage Spike During Soft-Switch Based on the Method of PI in Photovoltaic Converter System

For the problem of voltage spike which is caused by the MOSFET, and the issue of Variable calculation complex, given in the fly-back photovoltaic converter system (hereinafter abbreviated as PV). A new physical circuit of soft-switch control strategy is proposed which based on PI controller with the algorithm of variable power step to suppress the voltage spike value. According to the theoretical analysis, in this paper, the output voltage of DC part is equal to the input voltage of photovoltaic array and the reflected voltage of secondary side by clamping the voltage of clamped capacitor. The output voltage value from PI controller is used to recovery the leak inductance energy which will help the converter for working on the maximum power point(MPP) steadily. The new design can absorb the leak inductance energy and suppress the voltage spike efficiently. In this paper, for solve the traditional problem, we correct the formula of soft-switch by the method of maximum power point tracking (MPPT) algorithm. The new soft-switch algorithm formula can simplify the demand of physical circuit design through the method of variable power step.

Huixiang Xu, Nianqiang Li

A Novel Feature Extraction Method for Epileptic Seizure Detection Based on the Degree Centrality of Complex Network and SVM

Epilepsy is a kind of ancient disease, which is affecting the life of patients. With the increasing of incidence of epilepsy, automatic epileptic seizure detection with high performance is of great clinical significance. In order to improve the efficiency of epilepsy diagnosis, a novel feature extraction method for epileptic EEG signal based on the statistical property of the complex network and an epileptic seizure detection algorithm, which is composed of the extracted feature and support vector machine (SVM) is proposed. The EEG signal is converted to complex network by horizontal visibility graph firstly. Then the degree centrality of complex network as a novel feature is calculated. At last, the extracted feature and SVM construct automatic epileptic seizure detection. A classification experiment of the epileptic EEG dataset is performed to evaluate the performance of the proposed detection algorithm. Experimental results show the novel feature we extracted can distinguish ictal EEG from interictal EEG clearly and the proposed detection algorithm achieves high classification accuracy which can be up to 93.92 %.

Haihong Liu, Qingfang Meng, Qiang Zhang, Zaiguo Zhang, Dong Wang

An Improved Rife Algorithm of Frequency Estimation for Frequency-Hopping Signal

In this paper we analysis the performance of Rife algorithm and point out that the performance is poor when the true frequency is much close to quantized frequency of FFT. Aiming at this problem, an improved Rife algorithm is presented. This algorithm can make the signal frequency always located in the center between the two neighboring discrete frequencies by using zoom FFT technique. Then, in order to eliminate the impact of noise for the Rife algorithm, this paper propose to correct the improved Rife algorithm by using the window function for the signal. Simulation result showed that the improved algorithm had high estimation accuracy, strong anti-noise performance and good stability.

Jun Lv, Leying Yun, Tong Li

Blind Hyperspectral Unmixing Using Deep-Independent Information

In linear mixing model (LMM), the endmember fractional abundances should satisfy the sum-to-one constraint, which makes the well-known independent component analysis (ICA) based blind source separation (BSS) algorithms not well suited to blind hyperspectral unmixing (bHU) problem. A novel framework for bHU consulting dependent component analysis (DCA) is presented in this paper. By using the idea of subband decomposition, wavelet packet decomposition based bHU algorithm (termed as SDWP-bHU) is proposed, where the deep independent information of the source signals is exploited to fulfill the endmember signatures extraction and abundances separation tasks. Experiments based on the synthetic data are performed to evaluate the validity of the proposed approach.

Fasong Wang, Rui Li, Jiankang Zhang, Li Jiang

Speech Denoising Based on Sparse Representation Algorithm

A new speech denoising method that aims for processing corrupted speech signal which is based on K-SVD sparse representation algorithm is proposed in this paper. Here, the DCT sparse and redundant representation over dictionary is used for the initial redundant dictionary. In order to analyze the time-frequency characteristics of speech signal clearly, the spectrogram patches are applied as training samples for the sparse decomposition in this approach. However, the training samples need to extend their deployment to arbitrary spectrogram sizes because the K-SVD algorithm is limited in handling small size spectrogram. A global spectrogram was defined prior that forces sparsity over patches in every location in the spectrogram. Afterwards, by using the K-SVD algorithm, the greedy algorithm is used for updating which alternates between dictionary and sparse coefficients. Then a dictionary that describes the speech structure effectively can be obtained. Finally, the corrupted speech signal can be sparsely decomposed under the redundant dictionary. Consequently, the sparse coefficients can be obtained and used to reconstruct the noiseless spectrograms. As a result, the purpose of the separation for the signal and noise is reached. The proposed K-SVD algorithm is a simple and effective algorithm, which is suitable for processing corrupted speech signal. Simulation experiments show that the performance of the proposed K-SVD denoising algorithm is stable, and the white noise can be effectively separated. In addition, the algorithm performance surpasses the redundant DCT dictionary method and Gabor dictionary method. In a word, K-SVD algorithm leads to an alternative and novel denoising method for speech signals.

Yan Zhou, Heming Zhao, Xueqin Chen, Tao Liu, Di Wu, Li Shang

A New Method for Yielding a Database of Hybrid Location Fingerprints

Location fingerprinting in wireless LAN positioning used by RSSI (Received Signal Strength Indication) has become one of the most popular indoor positioning technologies. However RSSI is easily affected by environment, it is difficult to improve the accuracy. In this paper, a hybrid fingerprints construction algorithm was presented by RSSI and geomagnetic field which can effectively reduce the inaccuracy compared to the construction location fingerprints that only use RSSI. This article has also presented the C-GPCA (Clustering Grouped Principal Component Analysis) algorithm, which can be used in the further optimization of hybrid fingerprint database, improving the positioning accuracy.

Yan-Hua Li, Wen-Sheng Tang, Sheng-Chun Wang, Peng Hui

A Note on the Guarantees of Total Variation Minimization

In this paper, we provide a simplified understanding of the guarantees of 1-dimension total variation minimization. We consider a slightly modified total variation minimization rather than the original one. The modified model can be transformed into an $$ \ell_{1} $$ minimization problem by several provable mathematical tools. With the techniques developed in random sampling theory, some estimates relative to Gaussian mean width are provided for both Gaussian and sub-Gaussian sampling. We also present a sufficient condition for the exact recovery under Gaussian sampling.

Hao Jiang, Tao Sun, Pei-Bing Du, Sheng-Guo Li, Chun-Jiang Li, Li-Zhi Cheng

CDN Strategy Adjustment System Based on AHP

It has been an important research orientation on how to deploy the service strategy intelligently to meet the needs of complicated application scenarios for CDN carriers in order to improve its quality of service (QoS). In order to overcome the deficiency of traditional Static deployment strategy, this paper proposes a CDN Dynamic deployment strategy system based on Analytical Hierarchy Process (AHP). The system will try to structure an open hierarchical architecture and decide the weight of elements influencing the QoS of CDN, and will finally give out the recommendatory solution. It has been proved by the example in the paper that this decision model is efficient in deciding the proposed solution, which is objective and effective.

Xi Chen, Xie Zhang, Zongze Wu, Youjun Xiang, Shengli Xie, Shuang Li

Pattern Recognition

Frontmatter

Detection of Abnormal Event in Complex Situations Using Strong Classifier Based on BP Adaboost

In order to recognize the abnormal event, such as emergency or panic, happened in public scenes timely, an algorithm based on features extraction and BP Adaboost to detect abnormal frame event from surveillance video of complex situation is proposed. The proposed method detects an abnormal event where people are running, and this panic situation is simulated by the frame in a video. Experiments show that the method can distinguish and detect the abnormal event effectively and efficiently, which has potentiality to be used in the real public monitoring.

Yuqi Zhang, Tian Wang, Meina Qiao, Aichun Zhu, Ce Li, Hichem Snoussi

A Comparative Study for the Effects of Noise on Illumination Invariant Face Recognition Algorithms

In the literature, the effects of noise on existing face recognition algorithms are neglected, to the best of our knowledge. In this paper, for the first time, we perform an experimental study for the effects of noise on existing illumination-invariant face recognition algorithms. In total, twenty-one algorithms have been included in this study in this paper. We find that, when noise is present in face images, Tan and Triggs’ method achieves the highest correct recognition rates for both the extended Yale B face database and the CMU-PIE face database. If face images do not contain noise, isotropic smoothing is preferred because this method obtains the highest average recognition rate (96 %) for the extended Yale B database and 16 out of 21 methods achieve 100 % correct recognition rates for the CMU-PIE face database.

Guangyi Chen, Wenfang Xie

Algorithms of the Cluster and Morphological Analysis for Mineral Rocks Recognition in the Mining Industry

This paper describes an algorithm for automatic segmentation of color images of various ore types, using the methods of morphological and cluster analysis. There are some examples illustrating the usage of the algorithm to solve mineral recognition problems. The effectiveness of the proposed method lies in the area of automatic objects of interest identification inside the image, tuning the parameters of the amount allocated to the segments. This paper contains short description of morphological and cluster analysis algorithms for the mineral recognition in the mining industry.

Olga E. Baklanova, Mikhail A. Baklanov

A Similarity-Based Approach for Shape Classification Using Region Decomposition

Measuring the similarity of two shapes is an important task in human vision systems in order to either recognize or classify the objects. For obtaining reliable results, a high discriminative shape descriptor should be extracted by considering both global and local information of the shape. Taking into account, this work introduces a centroid-based tree-structured (CENTREES) shape descriptor invariant to rotation and scale. Extracting the CENTREES descriptor is started by computing the central of mass of a binary shape, assigned as the root node of tree. The entire shape is then decomposed into b sub-shapes by voting each pixel point according to an angle between point and major principal axis relative to a centroid. In the same way, the central of mass of the sub-shapes are calculated and these locations are considered as level-1 nodes. These processes are repeated for a predetermined number of levels. For each node corresponding to sub-shapes, parameters invariant to translation, rotation and scale are extracted. A vector of all parameters is considered as descriptor. A feature-based template matching with X2 distance function is used to measure shape dissimilarity. The evaluation of our descriptor is conducted using MPEG-7 dataset. The results justify that the CENTREES is one of reliable shape descriptors for shape similarity.

Wahyono, Laksono Kurnianggoro, Yu Yang, Kang-Hyun Jo

Comparison of Non-negative Matrix Factorization Methods for Clustering Genomic Data

Non-negative matrix factorization (NMF) is a useful method of data dimensionality reduction and has been widely used in many fields, such as pattern recognition and data mining. Compared with other traditional methods, it has unique advantages. And more and more improved NMF methods have been provided in recent years and all of these methods have merits and demerits when used in different applications. Clustering based on NMF methods is a common way to reflect the properties of methods. While there are no special comparisons of clustering experiments based on NMF methods on genomic data. In this paper, we analyze the characteristics of basic NMF and its classical variant methods. Moreover, we show the clustering results based on the coefficient matrix decomposed by NMF methods on the genomic datasets. We also compare the clustering accuracies and the cost of time of these methods.

Mi-Xiao Hou, Ying-Lian Gao, Jin-Xing Liu, Jun-Liang Shang, Chun-Hou Zheng

Deep Learning with PCANet for Human Age Estimation

Human age, as an important personal feature, has attracted great attention. Age estimation has also been considered as complex problem, how to get distinct age trait is important. In this paper, we investigate deep learning techniques for age estimation based on the PCANet, name DLPCANet. A new framework for age feature extraction based on the DLPCANet model. Different from the traditional deep learning network, we use PCA (Principal Component Analysis, PCA) algorithmic to get the filter kernels of convolutional layer instead of SGD (Stochastic Gradient Descent, SGD). Therefore, the model parameters are significantly reduced and training time is shorter. Once final feature has been fetched, we K-SVR (kernel function Support Vector Regression, K-SVR) for age estimation. The experiments are conducted in two public face aging database FG-NET and MORPH, experiments show the comparative performance in age estimation tasks against state-of-the-art approaches. In addition, the proposed method reported 4.66 and 4.72 for MAE (Mean Absolute Error, MAE) for point age estimation using FG-NET and MORPH, respectively.

DePeng Zheng, JiXiang Du, WenTao Fan, Jing Wang, ChuanMin Zhai

Natural Scene Digit Classification Using Convolutional Neural Networks

We used a convolutional neural networks based model to classify scene digits. We proposed the Horizontal and Vertical Feature Block to extract feature from different fields of the input images, which is efficient and has fewer parameters. We introduced a multi-input strategy to add location information to our model, while the traditional methods only use a part of information from the source annotations. More importantly, we released a new dataset for scene digit classification. The new dataset is collected from Baidu street view and mobile photos. The samples in the dataset are from the real world, and they are collected from many kinds of scenes in our daily lives, so that this dataset has huge potential in many applications.

Ziqin Wang, Peilin Jiang, Xuetao Zhang, Fei Wang

Deep Learning and Shared Representation Space Learning Based Cross-Modal Multimedia Retrieval

An increasing number of different multimedia information, including text, voice, video and image, are used to describe the same semantic concept together on the Internet. This paper presents a new method to more efficiently cross-modal multimedia retrieval. Using image and text as an example, we learn the deep learning features of images by convolution neural networks, and learn the text features by a latent Dirichlet allocation model. Then map the two features spaces into a shared presentation space by a probability model in order that they are isomorphic. At last, we adopt centered correlation to measure the distance between them. The experimental results in the Wikipedia dataset show that our approach can achieve the state-of-the-art results.

Hui Zou, Ji-Xiang Du, Chuan-Min Zhai, Jing Wang

Leaf Classification Utilizing a Convolutional Neural Network with a Structure of Single Connected Layer

Plant plays an important role in human life, so it is necessary to build an automatic system for recognizing plant. Leaf classification has become a research focus for twenty years. In this paper, we propose a single connected layer (SCL) structure adding into the convolutional neural network (CNN). We use this CNN model for plant leaf identification and report the promising results on ICL leaf database. Moreover, we propose some improvement on it to let it perform better. The result shows that our advanced SCL can effectively improve the accuracy of CNN.

Xiang He, Gang Wang, Xiao-Ping Zhang, Li Shang, Zhi-Kai Huang

Person Re-identification Based on Color and Texture Feature Fusion

Due to variations in pose, illumination condition, the appearance of a person differs significantly in two views, which makes person re-identification inherently difficult. In this paper, we propose a feature fusion method for person re-identification, which includes the HSV and color histogram features and the texture feature extracted by the HOG descriptor. The specific process is divided into training phase and recognition phase. In the training phase, we extract the feature descriptors of each image in the reference dataset, and then we learn a subspace in which the correlations of the reference data from different cameras are maximized using canonical correlation analysis (CCA). For re-identification, we extract the feature descriptors of each image in the gallery dataset and the probe dataset. And the feature descriptors are projected onto the CCA subspace and acquire the new feature descriptors. The re-identification is done by measuring the similarity between the gallery image descriptor and the probe image descriptor. Experimental results on standard benchmarking datasets show that the proposed method out performs the state-of-art approaches.

Li Yuan, Ziru Tian

Recognition of Mexican Sign Language from Frames in Video Sequences

The development of vision systems capable to extracting discriminative features that enhance the generalization power of a classifier is still a very challenging problem. In this paper, is presented a methodology to improve the classification performance of Mexican Sign Language (MSL). The proposed method explores some frames in video sequences for each sign. 743 features were extracted from these frames, and a genetic algorithm is employed to select a subset of sensitive features by removing the irrelevant features. The genetic algorithm permits to obtain the most discriminative features. Support Vector Machines (SVM) are used to classify signs based on these features. The experiments show that the proposed method can be successfully used to recognize the MSL with accuracy results individually above 97 % on average. The proposed feature extraction methodology and the GA used to extract the most discriminative features is a promising method to facilitate the communication of deaf people.

Jair Cervantes, Farid García-Lamont, Lisbeth Rodríguez-Mazahua, Arturo Yee Rendon, Asdrúbal López Chau

Robust Epileptic Seizure Classification

A lot of feature vectors and sub-band signals are considered for Epileptic seizure classification. Unfortunately, not all the feature vectors and sub-band signals contribute to the final result. In view of this limitation, we propose a modified Differential Evolution Feature Selection algorithm (MDEFS), which searches the best feature vector subset and the sub-band signals to distinguish three groups of subjects (healthy, ictal and interictal). From the experiment results, it is observed that the bagging method based on the optimal feature subset (the standard deviation attribute in the delta sub-band signal, the time-lag attribute in the delta sub-band signal, fractal dimension in the alpha sub-band signal, the correlation dimension attribute in the alpha sub-band signal and the standard deviation attribute in the beta sub-band signal) selected by MDEFS results in highest classification accuracy of 98.67 %.

Farrikh Alzami, Daxing Wang, Zhiwen Yu, Jane You, Hau-San Wong, Guoqiang Han

A Simple Review of Sparse Principal Components Analysis

Principal Component Analysis (PCA) is a common tool for dimensionality reduction and feature extraction, which has been applied in many fields, such as biology, medicine, machine learning and bioinformatics. But PCA has two obvious drawbacks: each principal component is line combination and loadings are non-zero which is hard to interpret. Sparse Principal Component Analysis (SPCA) was proposed to overcome these two disadvantages of PCA under the circumstances. This review paper will mainly focus on the research about SPCA, where the basic models of PCA and SPCA, various algorithms and extensions of SPCA are summarized. According to the difference of objective function and the constraint conditions, SPCA can be divided into three groups as it shown in Fig. 1. We also make a comparison among the different kind of sparse penalties. Besides, brief statements and other different classifications are summarized at last.Fig. 1.The classification of SPCA algorithms in this paper.

Chun-Mei Feng, Ying-Lian Gao, Jin-Xing Liu, Chun-Hou Zheng, Sheng-Jun Li, Dong Wang

Endpoint Detection and De-noising Method Based on Multi-resolution Spectrogram

The paper studied endpoint detection algorithm of noisy speech, since the visual differences of spectrogram employed by speech and noise, the paper chose spectrogram endpoint detection methods. Technical difficulties of spectrogram endpoint detection is how to describe the intuitive difference of spectrogram by mathematical amount, according to the descriptive power of autocorrelation coefficients on texture features, the paper described the difference by selecting the autocorrelation function, and proposed column autocorrelation spectrogram detection method. Through the distribution of spectrogram self-correlation function, as the threshold of endpoint detection for the noisy speech, the cut-off point between speech and noise was found out. Since the paper used broadband spectrogram, which employed poor frequency resolution, so there were still residual noise in speech column after autocorrelation spectrum detection, in order to further de-noising in different bands, combined with the multi resolution of empirical mode decomposition (EMD), the paper analyzed the noisy speech by multi-resolution, the target was broken down into different frequency scales and was further analyzed by column autocorrelation spectrogram, experiments shown that the noise reduction effect for noisy speech was ideal.

Jing Zhang

Biometrics Recognition

Frontmatter

An Efficient Face Recognition System with Liveness and Threat Detection for Smartphones

This paper proposes a face recognition system with liveness and threat detection on smartphone. Liveness and under-threat situations are decided through eye-blinking and facial expressions. It is designed to handle resource constraints of mobile devices such as low processing power, limited memory, less battery power and low quality of images. It uses Uniform Extended Local Ternary Pattern (UELTP) features for the threat detection. Whereas, Uniform Local Binary Pattern (ULBP) and Binarized Hamming Distance (BHD) are used for liveness detection. The experiments have been conducted on three in-house databases called SmartBioVideo, SmartBioFace and SmartBioThreatFace. Results have found to be promising and time efficient.

Kamlesh Tiwari, Suresh Kumar Choudhary, Phalguni Gupta

Image Processing

Frontmatter

Feature Extraction with Radon Transform for Block Matching and 3D Filtering

We propose a novel modification to patch matching in block matching and 3D filtering (BM3D), which is the state-of-the-art in image denoising. The BM3D calculates the distance between two patches by taking the sum of square of the pixel difference. However, when the noise level is very high, this patch matching technique will be less effective. It is well known that Radon transform is very good at suppressing Gaussian white noise and hence in this paper we use it to extract robust features from the two patches for patch matching in BM3D. Experimental results confirm the effectiveness of our proposed modification to BM3D for image denoising in heavily noisy scenarios.

Guang Yi Chen, Wen Fang Xie

A Novel Image Steganography Using Wavelet Contrast and Modulus Operation

Steganography is the science of hiding data into innocuous objects such that the existence of the hidden data remains imperceptible to an adversary. To obtain larger embedding capacity and imperceptible stegoimages, this paper proposes a new image steganography using wavelet contrast and modulus operation based on the visual characteristics that human are not sensitive to rapid change area and dark area. The method exploits the wavelet contrast value of image block to estimate how many bits will be embedded into the image block, and embeds secrete data by way of modulus operation. Our experimental results show that the proposed approach provides both larger embedding capacity and higher image quality.

Weiyi Wei, Yahong Wen

Efficient Specular Reflection Separation Based on Dark Channel Prior on Road Surface

On the busy section of road, vehicle collision could be easily caused by road surface reflection, because road surface reflection always leads to problems in stereo matching, recognition and segmentation. Few attentions have been paid to deal this problem till now. Existing methods rely on a specular free image to detect and estimate specular reflection. Their methods are suitable for image indoor rather than road surface image outdoor. Therefore they are not applicable to road surface reflection. In this paper, a novel method, called dark channel with threshold filter (DCTF) is presented to separate specular reflection from road surface. The method utilizes dark channel prior to roughly get an estimation of the road surface reflection. Then a threshold is proposed which can recover the specular reflection despite of the visual artifacts robustly. Experimental results show that our method significantly outperforms the previous methods in separating specular reflection on road surface.

Yao Wang, Fangfa Fu, Jinjin Shi, Weizhe Xu, Jinxiang Wang

The Scene Classification Method Based on Difference Vector in DCT Domain

Scene classification is one of the hot research topics in the field of computer vision, it is the basis of the organization and access for a variety of image database, so it has important practical significance. In our previous work, we put forward a novel fast scene classification method via DCT based on the energy concentration and multi-resolution characteristics of DCT coefficients. This paper improved our previous work proposed a scene classification method based on DCT domain using difference vectors. First of all, divided the whole image into the regular grid without repetition, in each grid, do DCT transform with the size of 8 * 8 get the DCT coefficients matrix, extract the AC coefficients in the matrix get the original vectors; Then, selected N images from each category in the database randomly, calculate the average vector of their original vectors, using the original vectors of all images corresponding category subtract the average vector get the difference vectors as the feature vectors; Finally, based on these feature vectors defined above, train classifiers with one-vs.-all support vector machine (SVM). In order to verify the robustness of the proposed algorithm, this paper has built an image database contains eight scene categories according to the OT database, this paper conducted cross validation experiment for the proposed method in the two databases. Experimental results show that the proposed method has higher accuracy and speed in image classification, and has good robustness.

Ce Li, Ming Li, Limei Xiao, Beijie Ren

Image Compression Based on Analysis Dictionary

Along with the extension of the application of the dictionary learned through the synthesis model in the image compression, the time consumption in the sparse representation becomes a key factor restricting the efficiency of the system. Therefore in view of the defect of the synthesis model in the application, combining with the advantages of the analysis model in the sparse representation, we proposed an image block compression model based on analysis dictionary (ALDBCS). In this model, a dictionary which is obtained by using the prior data, is introduced to the process of image compression. The reconstructed simulation experiment proves that the ALDBCS model can not only improve the quality of image reconstruction, but also reduce the consumption of image compression.

Zongwei Feng, Yanwen Chong, Weiling Zheng, Shaoming Pan, Yumei Guo

An Improved Algorithm Based on SURF for MR Infant Brain Image Registration

The correct diagnosis of brain diseases is crucial for children with brain disorders. But the complex characteristics of infant brain make the image analysis very complicated. Thus, an accurate image registration is a prerequisite for accurate analysis of MR infant brain images, and it provides valuable information for the diagnosis of doctors. This paper presents our research works on SURF registration algorithm of 2-D MR infant brain images. We firstly describe the original algorithm and analyze its advantages and drawbacks. Then an improved version is proposed, which uses 8-D descriptor vectors with the length of 128. The experiment results show, compared with the original version, our algorithm can achieve more accurate image registration with a little more time consumption. For all the images tested, the increase of correct matching rate varies from a minimum of 5.7 % to a maximum of 14.9 % compared with the classical one.

Ke Du, Stéphane Domas, Michel Lenczner, Guangjin Zhang

Slippage Estimation Using Sensor Fusion

In this paper, a non-contact slippage estimation approach using sensor fusion is proposed. The sensor consists of a charge-coupled device (CCD) camera and structured light emitter. The slip margin is obtained by estimating very small displacement of the grasped object in consecutive frames sequence captured by CCD camera. In experiments, we apply our approach on a slip-margin feedback control gripper system. The three degree of freedom (DOF) gripper consisting of a CCD camera, structured light and force sensor grasps a target object. The incipient slippage occurs on the contact surface between grip fingers and grasping object when the object is pressed and slid, is estimated by proposed approach. Then, the grip force is immediately controlled by a direct feedback of the estimated slip margin. Consequently, the force is adaptively maintained in order to prevent the object from damage. The proposed approach validity is confirmed by results of experiments.

Thi-Trang Tran, Cheolkeun Ha

K-SVD Based Image Denoising Method Using Image Residual Information in Different Frequency Bands

The common image denoising methods only consider how to restore well image information from noise images, but neglect the effects of residual information between restored images and given images. To enhance denoised image’s quality, a new image denoising method considering residual information in different frequency bands is discussed in this paper. In this method, an original image is divided into high and low frequency sub-band images by the contourlet transform algorithm. And each sub-band image is first denoised by the K-singular value decomposition (K-SVD) denoising model, thus each residual sub-band image is correspondingly obtained. Further, each residual image is again denoised by K-SVD denoising model. Finally, for each sub-band image denoised and its residual image, the inverse transform of contourlet transform algorithm is used to restore the original image. Compared our method proposed here with common denoising methods of wavelet, contourlet, K-SVD, experimental results show that our method fusing residual information in different frequency bands behaves better denoising effect.

Pin-gang Su, Tao Liu, Zhan-li Sun

Single Image Super Resolution with Neighbor Embedding and In-place Patch Matching

In this paper, we present a novel image super-resolution framework based on neighbor embedding, which belongs to the family of learning-based super-resolution methods. Instead of relying on extrinsic set of training images, image pairs are generated by learning self-similarities from the low-resolution input image itself. Furthermore, to improve the efficiency of image reconstruction, the in-place matching is introduced to the process of similar patches searching. The gradual magnification scheme is adopted to upscale the low-resolution image, and iterative back projection is used to reduce the reconstruction error at each step. Experimental results show that our method achieves satisfactory performance not only on reconstruction quality but also on time efficiency, as compared with other super-resolution methods.

Zhong-Qiu Zhao, Zhen-Wei Hao, Run Su, Xindong Wu

A Modified Non-rigid ICP Algorithm for Registration of Chromosome Images

As an extension of the classic rigid registration algorithm-Iterative Closest Point (ICP) algorithm, this paper proposes a new non-rigid ICP algorithm to match two point sets. Each point in the data set is supposed to match to the model set via an affine transformation. The proposed registration model is built up with a regularization term based on their average affine transformation. For each iteration of our algorithm, firstly correspondences between two point sets are built by the nearest-point search. Then the non-rigid transform parameters between two correspondence point sets are estimated by the proposed method in the closed form. Finally the average affine transformation is updated. A set of challenging data including single and overlapping chromosome images are tested which have significant local non-rigid transformations. Experimental results demonstrate our algorithm has higher accuracy and faster rate of convergence than other algorithms.

Qian Kou, Yang Yang, Shaoyi Du, Shuang Luo, Dongge Cai

Locally Biased Discriminative Clustering Method for Interactive Image Segmentation

Interactive image segmentation is a form of semi-supervised segmentation method by using the user interactive information. It performed well than fully unsupervised segmentation methods. In this paper, we propose a novel interactive image segmentation method, in which a seed vector is used to represent the user scribbles. Then a soft similarity constraint is added to the discriminative clustering model. The soft constraint allows the user to tune the degree which the constraint is satisfied. With respect to the discriminative clustering model, the clustering result is not affected by the assumption to the distribution of the data, and it’s easy to add constraint to the clustering variable. The final optimization problem is convex, so it can reach global optimal solution. The proposed method is evaluated on benchmark dataset BSD dataset, and it performs well than state of art methods both in quantitative and qualitative results.

Xianpeng Liang, Xiao-Ping Zhang, Li Shang, Zhi-Kai Huang

Accurate Prior Modeling in the Locally Adaptive Window-Based Wavelet Denoising

The locally adaptive window-based (LAW) denoising method has been extensively studied in literature for its simplicity and effectiveness. However, our statistical analysis performed on its prior estimation reveals that the prior is not estimated properly. In this paper, a novel maximum likelihood prior modeling method is proposed for better characterization of the local variance distribution. Goodness of fit results shows that our proposed prior estimation method can improve the model accuracy. A modified LAW denoising algorithm is then proposed based on the new prior. Image denoising experimental results demonstrate that the proposed method can significantly improve the performance in terms of both peak signal-to noise ratio (PSNR) and visual quality, while maintain a low computation.

Yun-Xia Liu, Yang Yang, Ngai-Fong Law

A Data Fusion-Based Framework for Image Segmentation Evaluation

Image segmentation is an important task in image processing. Nevertheless, there is still no generally accepted quality measure for evaluating the performance of various segmentation algorithms or even different parameterizations of the same algorithm. In this paper, we propose a data fusion-based binary classification framework for image segmentation evaluation. We train and test this framework using a dataset consisting of a variety of image types, their segmentations and respective ground truths, as well as the class labels assigned to each segmentation by human judges. Experimental results show accuracy of up to 80 %.

Macmillan Simfukwe, Bo Peng, Tianrui Li

Error Based Nyström Spectral Clustering Image Segmentation

Spectral clustering algorithm has been a research hotspot in the field of image processing, recent years. Spectral clustering based on the similarity of data while structure of similarity matrix is complex. The calculation of spectral clustering can be very time-consuming, especially in the process of Eigen-decomposition for Laplacian matrix. Nyström extension method could obtain the approximation solution of eigenvectors by using a small amount of sample information, reduce the computational complexity of spectral clustering effectively. Based on the features of image and the error analysis of Nyström a new sampling method is presented. Using Uniform Sampling generates a set of cluster centers at first; then, minimize the error between data and centers by iteration; finally, typical experiment results and analysis are given.

Liu Zhongmin, Li Bohao, Li Zhanming, Hu Wenjin

Supervised Online Dictionary Learning for Image Separation Using OMP

In this paper, we propose a new algorithm to perform single image separation based on online dictionary learning and orthogonal matching pursuit (OMP). This method consists of two separate processes: dictionary training for representing morphologically different components and the separation stage. The training process takes advantage of the prior knowledge of the components by adding component recovery error control penalties. The learned dictionaries have lower coherence with each other and better separation ability, which can benefit the separation process in two ways. Firstly, simple sparse coding methods such as OMP can be used to efficiently obtain superior performance. Secondly, well trained dictionaries can lead to satisfactory separation results even when the components are similar. The dictionaries obtained can also serve as good initial inputs for other models using dictionary learning and sparse representation. Experiments on complex images confirm that better results can be achieved efficiently by our method compared to other state-of-the-art algorithms.

Yuxin Zhang, Bo Yuan

Online Background-Subtraction with Motion Compensation for Freely Moving Camera

This paper proposes a background subtraction method for moving camera. The method relies on motion compensation to transfers the background model from the previous frame to the current frame. This motion compensation is carried out using homography transformation where the homography matrix is estimated from the set of point correspondences between previous and current frame. In order to achieve a fast processing speed, optical-flows from grid-based key-points are calculated to define the point correspondences. The background segmentation itself consists of 3 components: background model, candidate background model, and candidate age. Those 3 parameters are used to define the stable pixels which are considered as the background pixels. The proposed method was tested on a public benchmark system and achieved promising result as shown in the experimental report. Moreover, the method is able to work on real time with 56 fps of processing speed.

Laksono Kurnianggoro, Wahyono, Yang Yu, Danilo Caceres Hernandez, Kang-Hyun Jo

Computing the Number of Groups for Color Image Segmentation Using Competitive Neural Networks and Fuzzy C-Means

Fuzzy C-means (FCM) is one of the most often techniques employed for color image segmentation; the drawback with this technique is the number of clusters the data, pixels’ colors, is grouped must be defined a priori. In this paper we present an approach to compute the number of clusters automatically. A competitive neural network (CNN) and a self-organizing map (SOM) are trained with chromaticity samples of different colors; the neural networks process each pixel of the image to segment, where the activation occurrences of each neuron are collected in a histogram. The number of clusters is set by computing the number of the most activated neurons. The number of clusters is adjusted by comparing the similitude of colors. We show successful segmentation results obtained using images of the Berkeley segmentation database by training only one time the CNN and SOM, using only chromaticity data.

Farid García-Lamont, Jair Cervantes, Sergio Ruiz, Asdrúbal López-Chau

Improved Parallel Gaussian Elimination Algorithm in Magnetotelluric Occam’s Inversion

An improved parallel Gauss algorithm is put forward in MT Occam. Through analysing the process of the triangle, the eliminations of coefficient matrix and column matrix are merged. To avoid repeated calculation, column matrix elimination uses the intermediate result of coefficient matrix calculation directly. By defining two parameters, back substitution can use the result of coefficient matrix immediately. Meanwhile, the elimination triangle is divided to make the algorithm accord with the threads limit of the device. By using OpenCL the improved algorithm is implemented and applied to more different platforms. The experiments use two models under TE mode and TM mode to analysis the speedup. The results reveal that the improved algorithm can achieve higher speedup with solving large coefficient matrix size. Because air layer is added in TE mode, the coefficient matrix band expands, the triangle elements increase, and the speedup rises substantially. But the initialization time will account for a large proportion when solving smaller matrix size.

Yi Xiao, Pengdong Gao, Yongquan Lu

Extraction of Feature Points on 3D Meshes Through Data Gravitation

Feature points are particularly simple elements which demonstrate a model efficiently and availably; nonetheless, the points on 3D models cannot be extracted completely yet. Therefore, we propose a new algorithm based on data gravitation to extract the feature points on 3D meshes. First, we select the point with the maximum Gaussian curvature as the initial feature point set. Then, we use farthest point sampling to calculate the farthest distance from feature point set, and add this point into feature point set. Next we use the farthest distance to calculate data gravitation and select the point with the largest data gravitation until the farthest distance is smaller than a given threshold. Finally we get the feature points set on 3D meshes. In our experiments, we compare our algorithm with other algorithms. Results show that our algorithm can capture feature points effectively; consequently, the set of feature points reflects the features of 3D meshes precisely. Moreover, our algorithm is simple and is therefore easy to implement.

Chengwei Wang, Dan Kang, Xiuyang Zhao, Lizhi Peng, Caiming Zhang

An Improved Ultrasound Image Segmentation Algorithm for Cattle Follicle Based on Markov Random Field Model

In this paper, we proposed an improved ultrasound image segmentation algorithm for cattle follicle based on Markov random field model. According to the original ultrasound image dataset, we removed the speckle noise in ultrasound images by anisotropic diffusion filtering algorithm on the first step, and used the image enhancement technology to enhance the contrast of target area, then combined with an improved k-means algorithm for initial segmentation to realize basic classification of image pixels. As for the discontinuous over segmentation, we used area rule to remove the discontinuous over-segmentation region. Compared to the traditional MRF algorithm, this new algorithm has more accurate segmentation of the target area, better segmentation effect. The improved k-means algorithm to make initial segmentation for MRF model can also avoid initial clustering center to be selected randomly in comparison with the traditional k-means algorithm.

Jun Liu, Bo Guan

Three-Dimensional Cement Microstructure Texture Synthesis Based on CUDA

Three-dimensional reconstruction of cement is becoming increasingly important in cement hydration. Although many physical experiments have been conducted on cement hydration, and various algorithms have been developed to simulate the cement hydration for a long term, few algorithms have been developed to synthesize the three-dimensional microstructure of cement. Thus, we improve the Tree-structure Vector Quantization algorithm, which is effective in Unified Device Architecture. Experimental results indicate that the synthesis process in the proposed method is shorter and easier to implement which providing the same outcomes.

Kun Tang, Bo Yang, Lin Wang, Xiuyang Zhao, Yueqi Wang, Haixiao Zhang

Multiphase Image Segmentation Based on Improved LBF Model

In view of the problem of low efficiency of image segmentation with intensity inhomogeneity and the problem of the multi object image can’t be segmented, a new multi-phase image segmentation algorithm based on HLBF model is proposed. The application of magnetic resonance imaging in medicine is used to demonstrate the validity of the model. The proposed model replaces the Gauss kernel function in the original LBF model with the new kernel function to improve the time efficiency. Meanwhile, the HLBF model is further integrated into the variational level set of multi-phase image segmentation strategy to achieve the segmentation of multi-phase image with intensity inhomogeneity. Experimental results show the efficiency of the proposed method. The proposed model has advantages over the traditional segmentation method in terms of time efficiency and accuracy.

Ji Zhao, Huibin Wang, Han Liu

Multi-scale Spectrum Visual Saliency Perception via Hypercomplex DCT

Based on the salient object of human visual perception inconsistent scale, this paper proposes a multi-scale spectrum visual saliency perception with hypercomplex discrete cosine transform. In hypercomplex image color space to build parallel computing model, the method use HDCT to extract local spectral feature in an image. Meanwhile, the sparse energy spectrum calculated by hypercomplex discrete cosine transform on local image region was taken as visual stimulation signal. Then a visual saliency measurement was taken on both this region and its neighbor regions. Finally, the multi-sacle normalization was on the visual saliency response. The subjective and objective experimental results on the public saliency perception datasets demonstrated that both the precision and time cost of the proposed approach were better than the other state-of the-art approaches.

Limei Xiao, Ce Li, Zhijia Hu, Zhengrong Pan

Information Security

Frontmatter

An Efficient Conjunctive Keyword Searchable Encryption Scheme for Mobile Cloud Computing

The integration of the cloud and mobile device enables users more convenient to access, retrieve the file, but due to the limitations of resources of mobile devices, how to shorten the search time and get more accurate target files, to avoid unnecessary consumption has become the focus of research. This paper presents an efficient searchable encryption scheme based on mobile cloud, the scheme combine with k nearest neighbor algorithm, design the initial trapdoor matching table (TMT), realizes the multi-keyword Boolean search, improve the query precision, shorten the searching time.

Tao Lin, Zexian Sun, Hexu Sun, Bin Cao

Improvement of KMRNG Using n-Pendulum

The report mostly concentrates on ‘Keyboard Mouse Random Number Generator (KMRNG)’, which is made to overcome the limit of TRNG/PRNG and also including advantages of it being easy to be commercialized, by using Keyboard and Mouse inputs. Comparing SFMT (PRNG) and KMRNG with TestU01 showed us that even though KMRNG was proven suitable for random number generator, its statistical randomness was lower than that of SFMT. This also let us improve the KMRNG’s statistical performance by applying the statistical characteristic of the multiple pendulum.

Jae Jun Lee, Sungyoung Lee, Taeseon Yoon

XACML Policy Optimization Algorithm Based on Venn Diagram

This paper proposes an XACML (Extensible Access Control Markup Language) policy optimization algorithm to increase the efficiency of policy evaluation, which is based on the Venn graphic method of set theory. A three layer structure model for XACML is constructed. The policies and rules in the layers are mapped into sets and expressed with the Venn diagrams. According to the decision result of each layer and by setting the combining algorithm priority, the conflicts and the redundancies among access control policies and rules are detected and eliminated based on the intersection and union relations between sets. Experimental tests carried under the main evaluation engines show that the algorithm can decrease the evaluation time effectively and reduce the memory space occupancy as well.

Qiuru Lu, Jianping Chen, Haiying Ma, Weixu Chen

Virtual Reality and Human-Computer Interaction

Frontmatter

Usability Evaluation of the Flight Simulator’s Human-Computer Interaction

Attention is concentrated on the usability of the flight simulator’s Human-Computer Interaction system, especially on the cockpit display systems. Based on the User-Centered Design tool, used the cloud model to evaluate the usability of the cockpit display system from learnability, efficiency, memorability, errors and satisfaction. Cloud model based on grey system theory and the normal grey whitenization weight function used in conversion between qualitative and quantitative uncertainty model.

Yanbin Shi, Dantong Ouyang

Healthcare Informatics Theory and Methods

Frontmatter

Examining the Adoption and Use of Personally Controlled Electronic Health Record (PCEHR) System Among Australian Consumers: A Preliminary Study

This study looks at the adoption and use of Personally Controlled Electronic Health Record (PCEHR) system among consumers (individual users) in Australia. The specific aim of this study is to examine the current status of adoption and continued use of the PCEHR system among consumers in Australia. An online questionnaire survey was conducted, and 110 valid responses were received. The results of this study could contribute to the success of the ongoing roll-out of the PCEHR system in Australia and further studies in adoption and contused use of the system.

Jun Xu, Xiangzhu Gao, Golam Sorwar, Nicky Antonius

Artificial Bee Colony Algorithms

Frontmatter

Improved Artificial Bee Colony Algorithm Based on Reinforcement Learning

In order to overcome the basic artificial bee colony algorithm converges slowly and prematurely, the reinforcement learning is added into the artificial bee colony algorithm, in which several different updating strategies is mapped into an action used to update the nectar source location. According to the calculation of Q function value, each nectar source selects the optimal updating strategy to speed up the convergence rate. At the same time, the selection probability based on ranking is used instead of roulette wheel selection probability to keep population diversity and avoid premature convergence. Comparing with several different algorithms through the test functions and the parameter identification of Chaotic system. The results show that the proposed algorithm has higher accuracy and faster convergence rate, the feasibility and effectiveness of the algorithm is validated.

Ping Ma, Hong-Li Zhang

Differential Evolution

Frontmatter

Detect Method of Time Series’ Abnormal Value for Predictive Model

Abnormal value that in the predictive model of Ad Hoc networks may affecting the whole system’s working efficiency. We proposed a new detect method to dealing with this problem, constructed a forwarding model firstly, and then constructed a suitable model function through smoothing and modeling the time series. By using the mean shift model, we calculated the time series posterior probabilities and abnormal perturbation values, and then adjusted them, so as to weakening the influence of time series abnormal value. To verifying the efficiency of this forwarding method, we selected a time series with 300 observation points as the numerical example, statistics and analysis results indicate that it will be helpful to improve the efficiency of prediction models if we using this method.

Yang Feng

Memetic Algorithms

Frontmatter

Solving Bi-objective Unconstrained Binary Quadratic Programming Problem with Multi-objective Backbone Guided Search Algorithm

This paper presents a multi-objective backbone guided search algorithm in order to optimize a bi-objective unconstrained binary quadratic programming problem. Our proposed algorithm consists of two main procedures which are hypervolume-based local search and backbone guided search. When the hypervolume-based local search procedure can not improve the Pareto approximation set any more, the backbone guided search procedure is applied for further improvements. Experimental results show that the proposed algorithm is very effective compared with the original multi-objective optimization algorithms.

Li-Yuan Xue, Rong-Qiang Zeng, Yang Wang, Ming-Sheng Shang

Swarm Intelligence and Optimization

Frontmatter

Discrete Chaotic Gravitational Search Algorithm for Unit Commitment Problem

This paper presents a discrete chaotic gravitational search algorithm (DCGSA) to solve the unit commitment (UC) problem. Gravitational search algorithm (GSA) has been applied to a wide scope of global optimization problems. However, GSA still suffers from the inherent disadvantages of trapping in local minima and the slow convergence rates. The UC problem is a discrete optimization problem and the original GSA and chaos which belong in the realm of continuous space cannot be applied directly. Thus in this paper a data discretization method is implemented after the population initialization to make the improved algorithm available for coping with discrete variables. Two chaotic systems, including logistic map and piece wise linear chaotic map, are used to generate chaotic sequences and to perform local search. The simulation was carried out on small-scale UC problem with six-unit system and ten-unit system. Simulation results show lower fuel cost than other methods such as quadratic model, selective pruning method and iterative linear algorithm, confirming the potential and effectiveness of the proposed DCGSA for the UC problem.

Sheng Li, Tao Jiang, Huiqin Chen, Dongmei Shen, Yuki Todo, Shangce Gao

A Discrete Biogeography-Based Optimization for Solving Tomato Planting Planning

The yield of tomato affects the processing ability of ketchup factory directly. To improve the imbalance supply of the materials during tomato sauce season, building the mathematical model of tomato planting planning, a discrete Biogeography-based Optimization is proposed for solving tomato planting planning model. Considering the tomato planting planning is a large-scale combinatorial optimization problem, tomato planting matrix can be compressed by sparse matrix compression method to achieve compression of the solution space. And a new kind discrete BBO with a new coding way was used for planting planning. A tomato plant provides data in Xinjiang as an example of simulation calculation, the results showed that tomato planting planning scheme calculated by the proposed algorithm can realize the balance supplement of tomato materials effectively.

Hong-li Zhang, Cong Wang

A Multi-agent Approach for the Newsvendor Problem with Word-of-Mouth Marketing Strategies

Word-of-mouth (WOM) marketing is increasingly playing an important role in consumers’ purchase decision with the development of mobile Internet and various social media APP. We are particularly interested in such a problem as how to make decisions under effects of WOM campaigns? To answer this question, we develop a multi-agent model that emulates WOM or viral marketing process as spread of disease among people. Assume that each “infected” individual will purchase one unit of product. Then, the total “infected” people form the demand of the product, as an input of newsvendor problem. Besides finding the optimal order quantity of newsvendor problem, we also identify the most influential source node for kick off of the WOM marketing. The simulation results reveal that social network and WOM have a great influence on demand and profit of the firm. Even the source node has significant effect on output of WOM marketing. According to our simulation, the closeness centrality in social network analysis is the best measure to recognize the most influential source node, comparing to degree centrality, or betweenness centrality, etc. Finally, parameter analysis infers that profit of the firm will increase with higher the spreading probability or/and lower the resistant probability.

Feng Li, Ning Lin

Economic Dispatch of Grids Based on Intelligent Coordination Between Electric Vehicle and Photovoltaic Power

There are growing interests in electric vehicles because of environment concern. When a large number of EVs are introduced a power system, there will be extensive impacts on the security and economy of power system operation. Given this background, a dispatch optimization model is built in this paper to mitigate the peak-to-valley ratio of equivalent loads and reduce the active power losses of the distributed grid, based on the driving characteristics of EVs and other demands in power system. Particle swarm optimization algorithm is adopted to solve this optimization problem. The proposed model and algorithm are applied to the IEEE 33-bus test system, and the simulation results indicate the feasibility and efficiency.

Guangqing Bao, Weisheng Li, Dunwei Gong, Jiangwei Mao

Study on Tracking and Detecting Weak Multi-target Based on KF-GMPHDA in Multi-radar Networking

Gaussian mixture probability hypothesis density algorithm (GMPHDA), which is suitable for tracking weak signal to noise ratio (WSNR) multi-target, has rigorous theoretical foundation. The states and number of WSNR multi-target are tracked accurately by GMPHDA application in multi-radar networking, forming KF-GMPHDA. A suite of algorithm about KF-GMPHDA in multi-radar networking is proposed, improving track and detect algorithm in multi-radar networking. Simulation results show that all WSNR multi-target are tracked in multi-radar networking, which gets target tracks corresponding one to one with real targets by the proposed KF-GMPHDA. And then these guarantee higher-up to make full use of track information to acquire real targets states and judge battlefield.

Hai-Long Ding, Wen-Bo Zhao, Luo-Zheng Zhang

Study on Important Parameters of Tracking and Detecting RNWT Based on GMPHDA in Radar Networking

Gaussian mixture probability hypothesis density filter (PHDF), which is suitable for tracking RNWT (Targets in false alarm and clutter area such as stealth aircraft and UAV), has rigorous mathematical foundation. But the distribution covariance P and truncation threshold T of Gaussian elements in PHDF have not got reasonable calculation rules as yet, which bring bad influence to PHDF. Because the residual covariance S will be inverse calculated when it is involved in the gain calculation. If S is non-positive definite, it would lead to divergence calculation. To determine the P and T calculation rules we do probability statistics derivation. To solve the S calculation problem we do Cholesk and QR decomposition. The simulation compares demonstrate that PHDF in radar networking, using the proposed calculation rules of P, T and S, can precisely track RNWT multi-target, containing exist, birth and spawn targets, bring no extra calculation burden.

Hai-Long Ding, Wen-Bo Zhao, Guo-Chun Zhu

Tracking Number Time-Varying Nonlinear Targets Based on SQUF-GMPHDA in Radar Networking

For the problem that detecting and tracking targets in false alarm and clutter area (RNWT) is under constraints that motion model and measurement model are both nonlinear, firstly square root unscented filter algorithm (SQUFA) is used in nonlinear tracking in multi-radar networking. Then SQUFA is introduced to GMPHDA to form square root unscented filter Gaussian mixture probability hypothesis density filter algorithm (SQUF-GMPHDA), where newborn targets, spawn targets and existing targets are independently sampled, predicted and updated based on SQUFA. So the problem of high-precisely tracking RNWT under nonlinear condition is solved.

Hai-Long Ding, Wen-Bo Zhao, Luo-Zheng Zhang

Study on Tracking Strong Maneuvering Targets Based on IMM-GMPHDA

Gaussian mixture probability hypothesis density filter algorithm (GMPHDA), which is effective method for tracking unknown number of multi-target in strong clutter environment, has solid theoretical basis. But it is hard to track target by GMPHDA when the targets maneuver. To model maneuvering target, we introduce interacting multi-model (IMM) in GMPHDA by modeling maneuvering model of survival target and fusing probability hypothesis density of each model filter based on latest model probability, getting IMM-GMPHDA. The simulation results show that we can real-time track strong maneuvering and supersonic multi-target with IMM-GMPHDA, whose tracking precision can reach 70 m in multi-radar networking system, which meets the project requirement.

Hai-Long Ding, Wen-Bo Zhao, Luo-Zheng Zhang

Network Topology Management Optimization of Wireless Sensor Network (WSN)

Network topology management is one of the critical concerns when designing a Wireless Sensor Network (WSN). In this research, four basic factors including the total production cost, sensing coverage, network connectivity and fault tolerance are considered. A mathematical model is proposed to optimize four optimization metrics corresponding to the four design factors. This approach attaches a weighting coefficient to each optimization metric to adjust their importance in the optimization model. To solve the proposed model, an Ant Colony Optimization (ACO) based metaheuristics method, called MAX–MIN Ant System (MMAS) is used. In the experiment, Greedy algorithm (Greedy) and Genetic Algorithm (GA) are also adopted to solve the proposed model. The results indicate that MMAS shows a satisfactory performance on solving the proposed model, which there is an improvement on the number of sensor nodes comparing to the result of Greedy, and a better fitness value than the result of GA.

Chun Kit Ng, Chun Ho Wu, W. H. Ip, J. Zhang, G. T. S. Ho, C. Y. Chan

Soft Computing

Frontmatter

Tourism Network Comments Sentiment Analysis and Early Warning System Based on Ontology

Mine Tourism information and opinion, intelligent analysis user emotion, to improve tourism products and services, is the key to the success of tourism e-commerce. This paper embarks from the tourism network review information, researches how to build the microblog emotional vocabulary ontology and how to classify emotion based on Naive Bayes classification algorithm, implements a tourism network comments sentiment analysis and early warning system based on ontology. It not only save a large amount of manpower and material resources, but also have a certain reference value to establish reasonable tourism policy.

Yanxia Yang, Xiaoli Lin

Protein Structure and Function Prediction

Frontmatter

Prediction of Lysine Acetylation Sites Based on Neural Network

Lysine acetylation is a crucial type of protein post-translational modification, which is involved in many important cellular processes and serious diseases. In practice, identification of protein acetylated sites through traditional experiment methods is time-consuming and laborious. Computational methods are not suitable to identify a large number of acetylated sites quickly. Therefore, machine learning methods are still very valuable to accelerate lysine acetylated site finding. In this study, many biological characteristics of acetylated sites have been investigated, such as the amino acid sequence around the acetylated sites, the physicochemical property of the amino acids and the transition probability of adjacent amino acids. A special structure neural network, which is named flexible neural tree (FNT), was then utilized to integrate such information for generating a novel lysine acetylation prediction system named LA+FNT. When compared with existing methods, our proposed method overwhelms most of state-of-the-art methods. Such method has the ability to integrating different biological features to predict lysine acetylation with high accuracy.

Wenzheng Bao, Zhichao Jiang, Kyungsook Han, De-Shuang Huang

A Parallel Multiple K-Means Clustering and Application on Detect Near Native Model

Protein structure clustering is an important and essential step in protein 3D structure prediction. However, two issues limited current methods. But the large-scale candidate models in the decoy and undistinguished metric limit current methods to identify the near-native models. In this paper we proposed a novel method based on parallel multiple K-means cluster algorithms to identify the near-native structures. Parallel is introduced to reduce the memory and time consumption and multiple K-means to fusion different metrics of protein 3D similarity. Tested on 56 proteins, MK-means can well identify 33(58.9 %) proteins which are better or the same to SPICKER selected and 10 of the 33 proteins is the same results to the SPICKER. It indicates the performance of MK-means is similar to the top protein clustered tools SPICKER.

Hongjie Wu, Chuang Wu, Chen cheng, Longfei Song, Min Jiang

Computational Analysis of Similar Protein-DNA Complexes from Different Organisms to Understand Organism Specific Recognition

Protein-DNA interactions play vital roles in many cellular processes. It is not clear whether the recognition of same protein-DNA complexes in different organisms is similar or different. In this work, we have analyzed the similarities and variations in interactions and interacting patterns in a set of 41 similar protein-DNA complexes from different organisms based on several features such as propensity of binding site residues, preference of binding segments, preferred amino acid-nucleotide pairs, etc. Based on the analysis, we showed the variations in interactions and interacting patterns of similar protein-DNA complexes from different organisms, which possibly lead to difference in recognition mechanism.

R. Nagarajan, M. Michael Gromiha

Advances in Swarm Intelligence: Algorithms and Applications

Frontmatter

Adaptive Structure-Redesigned-Based Bacterial Foraging Optimization

This paper proposes an adaptive structure-redesigned-based bacterial foraging optimization called ASRBFO. In this improved algorithm, the chemotaxis step of SRBFO is adaptively adjusted based on the bacterial searching status. The personal current and best positions of bacteria as well as the mean of all bacterial positions are taken and used to calculate the chemotaxis step during the searching process. The goal of the study is to improve the convergence efficiency and the accuracy of SRBFO. To demonstrate the performance, six different benchmark functions are chosen to the experiment, and other three SRBFOs are used to compare with the proposed algorithm. The results show that ASRBFO outperforms other SRBFOs.

L. J. Tan, W. J. Yi, C. Yang, Y. Y. Feng

Artificial Bee Colony Optimization for Yard Truck Scheduling and Storage Allocation Problem

The yard truck scheduling (YTS) and the storage allocation problem (SAP) are two significant sub-issues in container terminal operations. This paper takes them as a whole optimization problem (YTS-SAP) and analyzes the factor of different travel speeds of trucks based on different loads. The goal is to minimize the total time cost of the summation of the delay of requests and the travel time of yard trucks. Due to the simplicity and easy implementation of artificial bee colony (ABC), the algorithm is applied to address the issue. Computational experiment is employed to examine and analyze the problem solutions and the performance of ABC algorithm. Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) are chosen as contrastive algorithms. From the results of the computational experiment, it is found that ABC algorithm can achieve better solution for the YTS-ASP problem.

Fangfang Zhang, Li Li, Jing Liu, Xianghua Chu

A Cooperative Structure-Redesigned-Based Bacterial Foraging Optimization with Guided and Stochastic Movements

The nested loop adopted in the original bacterial foraging optimization (BFO) is quite time-consuming and is the main reason for the complex computational process. Thus, in our previous work, an improved BFO with structure redesigned mechanism (SRBFO) is used to address this problem. Since the bacterial chemotaxis with stochastic direction in the original BFO has an adverse effect on the convergence rate, this paper proposes a new cooperative chemotactic movement strategy. In this cooperative strategy, some bacteria are selected to move toward a guided direction based on a predefined probability, while the other bacteria still swim to a stochastic direction as exhibited in the original BFO. By this strategy, all the bacteria alternatively use the guided movement and the stochastic movement to cooperatively balance global search and local search. The proposed improved algorithm is called Cooperative SRBFO (CSRBFO). A comparison of the CSRBFO with other BFOs has been made to demonstrate the superiority of the proposed algorithm.

Ben Niu, Jing Liu, Fangfang Zhang, Wenjie Yi

Backmatter

Weitere Informationen