Foundations and Practical Applications of Cognitive Systems and Information Processing

Mental rotation of hands, which is subject to biomechanical constrains, involves participants engaging in motor imagery processing. To contribute to a better understanding of the process of hand mental rotation, reaction times and event-related potential were measured while participants were performing a left–right hand recognition task. Participants apparently solved the task by imagining their own hands rotating to the orientation of the stimulus for comparison. In line with previous studies, the behavioral results showed that slower reaction times were found for the hand views that could not be easily reached with real movement. More importantly, the event-related potential results revealed that the amplitude of rotation-related negativity (RRN) decreased with the difficulty of the hand views increasing. The previous results are complemented by this study; it is stimulus views that modulate reaction times and the amplitude of RRN during mental rotation task of hands.

Predictive coding/biased competition (PC/BC) is a computational model of primary visual cortex (V1). Recent literature demonstrates that PC/BC model provides an implementation of the V1 bottom-up saliency map hypothesis. In this paper, we propose a novel approach toward natural color images saliency detection via the PC/BC model with top-down cortical feedback as context. We compare our method with the five state-of-the-art models of saliency detectors. Experimental results show that our method performs competitively for visual saliency detection task.

Recently, a growing number of neuroimaging studies have begun to pay attention to exploring the brains of schizophrenic patients to identify heritable biomarkers for this disorder involving their healthy siblings. Based on whole-brain resting-state functional connectivity of schizophrenic patients, their healthy siblings and healthy controls, the objective of the present study aimed to use multiclass pattern analysis to reveal three types of neural signature: (i) state connectivity patterns, reflecting the state of having schizophrenia; (ii) trait connectivity patterns, reflecting the genetic vulnerability to develop schizophrenia; and (iii) compensatory connectivity patterns, underlying special brain connections by which healthy siblings compensate for an increased genetic risk for developing schizophrenia. The current study may provide additional insights into the pathophysiological mechanisms underlying schizophrenia and be helpful in further highlighting genetic contribution to the etiology of schizophrenia.

In order to characterize non-random organization patterns of information process in the brain, we combine complex network analysis and resting-state functional magnetic resonance imaging to investigate brain activity derived from young adults, and then extract the tree layout and module structure of whole-brain network. These network organizations may be associated with the emergence of complex dynamics that supports the brain’s moment-to-moment responses to the external world and widen understanding potentially biological mechanisms of brain function.

For the real-time requirement of reconfigurable control allocation problem in the field of multi-surfaces aircraft, the control allocation scheme based on max direction derivative increment (MDDI) fixed point (FXP) iteration is proposed. The increment update for current iteration along the MDD and the design steps are given. Moreover, the convergence of the improved method is also proved. Comparisons of different methods are simulated in multi-surfaces aircraft model. The simulation results show the rapidity of MDDIFXP method compared with the original one and the effectiveness of the method in solving reconfigurable control allocation problem of multi-surfaces aircraft.

In intelligent video surveillance, multiple cameras, even a distributed network of video sensors, have to be employed to monitor activities over a complex area nowadays. Hence, the continuous object tracking across multiple cameras and object handover between adjacent cameras is urgently needed, in which many appearance cues and spatial–temporal information can be employed. This paper fuses the spatial–temporal cues with appearance cues into a particle filter to handle the camera handover with multiple cameras having non-overlapping view. The spatial–temporal cues, including source and sink regions, their transition probabilities, and transition time among adjacent regions, are learned offline. Then a spatial–temporal progressive matching scheme using particle filter is proposed to deal with camera handover among adjacent cameras. In particle filter matching course, the commonly used appearance cue, i.e. the histogram in HSV color space is used. Once an object enters into sink region, we first continuously scatter particles in source regions related to this sink region according spatial–temporal information until the object emergence detected, and secondly, based on the particle weights of every source region, adjust their particle numbers till the camera handover is successfully completed. Encouraging experiment results show the efficiency of this scheme.

An approach is presented based on driver simulator and SmarteyeII eye tracking system to examine the effects of pressing in-vehicle radio button on driver’s visual cognition. Parameters of glance frequency, glance duration, eye movement speed, and visual line moving in different regions of interest (ROIs) in task of pressing the radio button, closely related with driver’s visual cognition, were collected and analyzed. Based on the experimental data, driver’s visualization model with secondary tasks was built by CogTool. Driver’s vision, eye movement, cognition, and hand motion were tracked and recorded by the model. Results of experiment and running model show that pressing the in-vehicle radio button while driving has adverse influence on driver’s visual cognition and occupies a lot of the driver’s visual resources.

Guarantee to QoS for real-time traffic is the bottleneck to Ad hoc Networks’ development. In the distributed network, packet contention is so high that nodes always fail to access channels in limited time. Although there are many protocols, for example FPRP, which could eliminate collision remarkably, they usually need a large quantity of control packets, which lead to high access delay. By studying dynamical slot-distribution MAC protocol, a protocol using progressive reservation mechanism for Ad hoc Networks was laid out, named PR-MAC. In the new protocol, nodes access channel by gradation, coordinated competition, and also the idle data slot is reconstructed to resolve the confliction. Therefore, PR-MAC possesses the merit of high utilization of resources, few data-packets collision, and low access delay; in addition, it can also support data traffic fine.

This paper proposed an automatic steady-state visual evoked potential (SSVEP) component selection (SCS) measure for a high-performance SSVEP-based brain-computer interface (SBCI) system. First, multi-electrode raw electroencephalogram signals are spatially pre-processed using a blind source separation technique resulting in multi-source components. The SCS measure of each component is then calculated by continuous wavelet transform (CWT), and the ensemble features that contain the weighted CWT energy of individual SSVEP harmonic are extracted. Second, the SSVEP component with maximal SCS measure is considered to have the highest signal-to-noise ratio. In our SBCI system, six stimulus frequencies served as the input patterns. Offline analyses were performed, through which the common electrode locations, the time window size, and the number of harmonics were defined. Thereafter the results of our method were compared with those of others. We next carried out an online test of the SBCI for 11 subjects using eight common electrode locations, a 1.5-s time window, and the first and second harmonics. The test results showed that our method achieved an average accuracy of 95.2 % and a practical bit rate of 68.2 bits/min.

In this study, employing event-related potential (ERP) in response to faces and object stimuli, we explored the temporal course of cognitive biases and sex differences for facial attractiveness during a visual oddball paradigm. 10 women and 10 men were confronted with this task, within which they were asked to point out, as fast as possible, rare attractive or unattractive faces of neutral expression among a series of frequent stimuli (neutral objects). Behavioral analyses showed that men yielded longer reaction times than women, and deviant attractive faces were detected more slowly compared with deviant unattractive ones only for men. In accordance with the behavioral results, the ERP results showed that with respect to women, the N2b peak latencies were prolonged for both attractive and unattractive faces in men, perhaps reflecting early implicit attention to distinctive faces. Thereafter, for both sexes, deviant attractive faces evoked greater P3b amplitudes in comparison to deviant unattractive faces, revealing the cognitive biases toward facial beauty. Importantly, only in men, the P3b peak latencies were longer for attractive faces as opposed to their unattractive counterparts. Thus, it is likely that sex differences found in the detection of facial attractiveness could begin quite early in the information processing mechanism. Moreover, from an evolutionary view, the ERP and behavioral evidence together confirmed a reasonable supposition that although both men and women showed processing preferences for attractive faces, compared with women, men might attribute more value to distinctive evolution-related cues, especially to attractive information.

Particle filter (PF) is used in the three-dimensional (3D) free hand tracking system, which is nonlinear and non-Gaussian. Markov chain Monte Carlo (MCMC) plays a positive role in Bayesian statistical calculation and the maximum likelihood estimation. This paper focuses on using of MCMC algorithm in the PF sampling to reduce the time cost. The 3D free hand tracking system is real-time by using the improved PF algorithm. First, do experiments in the virtual platform with data gloves and establish constraints of 3D free hand. Second, we analyze the obtained data to get the sampling model, which is applied into the PF algorithm. Finally, use VC++ to code the algorithm in 3D gesture tracking system, and then compare with correlation algorithms. The results show that the cost of time is reduced by more than 15 % than the human gesture part recognition sample method (HGPRS) with the high tracking accuracy.

In cognitive radio (CR) networks, spectrum sensing which attracts a lot of interest is a significant task. Aiming at the problem that conventional spectrum sensing technique is usually focused on signal band. This paper introduces a multi-band joint spectrum detection based on multiple signal classification (MUSIC) algorithm for orthogonal frequency division multiplexing (OFDM) cognitive ratio system. The proposed eigenvalue-construct method only uses signal autocorrelation of OFDM symbols and simple sorting to achieve the spectrum detection. The computer simulations show that the proposed approach has a good performance compared with the conventional energy sensing method which uses the same threshold over multiple frequency bands.

Real-time target tracking is an important subject in modern intelligent surveillance and security defense systems. However, due to the natural scene’s complexity and variability, the tracking becomes complex and difficult especially when the target is occluded in complex background. This paper proposes a tracking algorithm of moving target based on adaptive blocking and feature correlation matching. We compute target’s grayscale first, and judge the target’s grayscale attribute. Then according to it, we choose a more suitable algorithm to track moving target from edge correlation matching algorithm and grayscale correlation matching algorithm based on multi-subblock. To edge matching algorithm, target’s displacement in two successive frames is determined by optimal matching of current unoccluded edge with real-time updated target template. For grayscale correlation matching based on multi-subblocks, the algorithm first estimates occluded region accurately by subblocks with distinct feature, and then tracks the target by residual unoccluded subblocks to participate in grayscale correlation matching. The experimental results of our tracking system show that the algorithm is effective for tracking moving targets.

An intuitive systemic model based on the systemic Yoyos and stochastic differential geometry is provided for finding a meaningful geometric description of radar-specific emitter identification (SEI) problems in this paper. According to this model, we show that intrinsic parameters of signals can be used to explain and find the effective fingerprints feature of specific emitters. Experiments on actual intercepted radar signals with the same type verify the correctness and validity of the proposed model.

This paper presents an implementation of indoor Simultaneous Localization and Mapping (SLAM) using RGBD images. Such system can be used in applications such as indoor robot navigation and environment perception. We perform coarse frame alignments using visual features. The coarse alignment results are then refined by applying Iterative Closest Point (ICP) algorithm to the point clouds. We create a pose graph which consists of keyframes which will be optimized if a new loop is detected. The performances of coarse alignment are tested using four methods—KLT tracker, SIFT, SURF and ORB. The experiment results show that ORB is a good trade-off between accuracy and efficiency. The performances and limitations of ICP are also explored. The results indicate that ICP is very sensitive to the initial value and the size of the point clouds. We also find that the loop closing largely reduces the alignment error. The maps of our laboratory are created using both the 3D point clouds and octomap.

Mobile maps, which integrates mobile positioning, mobile computing, and mobile communication technologies, differ from desktop and Web maps. Due to their small sizes, maintaining good legibility in mobile maps is very challenging. The variable-scale map model can be used in mobile maps to improve their legibility. Current variable-scale models often require human expertise to select appropriate parameters or models for generating variable-scale maps. However, these models ignore the effect of uneven distribution of spatial objects in map views. In this paper, a variable-scale model based on neighborhood is proposed, which selects the user’s focus region based on the neighborhood, identifies the focus region shape and automatically selects the variable-scale range and model. The principle and algorithm are introduced and described. Finally, the proposed model is evaluated. Results of the experiment show that the new model is feasible and valid to overcome the problem mentioned above.

A hybrid training algorithm named ODE-LM, in which the orthogonal differential evolution (ODE) algorithm is combined with the Levenberg-Marquardt (LM) method, is proposed to optimize feedforward neural network weights and biases. The ODE is first applied to globally optimize the network weights in a large space to some extent (the ODE will stop after a certain generation), and then LM is used to further learn until the maximum number of iterations is reached. The performance of ODE-LM has been evaluated on several benchmarks. The results demonstrate that ODE-LM is capable to overcome the slow training of traditional evolutionary neural network with lower learning error.

Most of vision based transport parameter detection algorithms are designed to be executed in good-natured weather conditions. However, limited visibility in rain or fog strongly influences detection results. To improve machine vision in adverse weather situations, a reliable weather conditions detection system is necessary as a ground base. In this article, a novel algorithm for weather condition automatic recognition is presented. This proposed system is able to distinguish between multiple weather situations based on the classification of single monocular color images without any additional assumptions or prior knowledge. Homogenous area is extracted form top to bottom in scene image. Inflection point information which implies visibility distance will be taken as a character feature for current weather recognition. Another four features: power spectral slope, edge gradient energy, contrast, saturation, and image noisy which descript image definition are extracted also. Our proposed image descriptor clearly outperforms existing descriptors for the task. Experimental results on real traffic images are characterized by high accuracy, efficiency, and versatility with respect to driver assistance systems.

Motion estimation is the most time consuming part in H.264. UMHexagonS algorithm was accepted as the fast motion search algorithm in H.264 reference software JM because of its short motion estimation time (MET) and good rate-distortion performance. In this paper, an improved algorithm based on UMHexagonS is proposed. Improved initial search point prediction, improved premature termination rule, octagon-diamond pattern, and new uneven multi-hexagon-grid search pattern are adopted. Experiment results showed that, it can reduce MET by at least 27.46 % compared with that of UMHexagonS without degrading video quality significantly.

A novel bionic, middle-semantic object annotation framework is presented in this paper. Moreover, we build the model based on the perception as defined by the human visual system. At first, the super-pixel is used to represent the images, and conditional random field could label each of the super-pixels, which means annotating the different classes of objects. In next step, on the basis of the previous result, image pyramid is used to represent the image, and get the sub-region of some objects of the same class. After extracting descriptor to represent the patches, all the patches are projected to a manifold, which could annotate the different views of objects from the same class. Experiments show that the bionic, middle-semantic object annotation framework could obtain superior results with respect to accuracy, and it could verify the correctness of WordNet indirectly.

A novel bionic, parallel, and hierarchical scene classification framework is presented in this paper. Moreover, we build the model based on the perception as defined by the human visual system. At first, we use an image pyramid to present both the global scene and local patches containing specific objects. Second, we build our own codebooks, which satisfy both long stare and short saccade similar to humans. Next, we train the visual words by generative and discriminative methods, respectively, which could obtain the initial scene categories based on the potential semantics using the bag-of-words model. Then, we use a neural network to simulate a human decision process. This leads to the final scene category. Experiments show that the parallel, hierarchical image representation, and classification model obtain superior results with respect to accuracy.

A kind of adaptive sub-channel allocation method utilizing Hopfield neural network (HNN) is studied in this paper. In order to find the power optimal sub-channel allocation under the constraints that only one sub-channel can be allocated to one user and all users are allocated the same number of sub-channels, a kind of new energy constrained function is constructed for the HNN. It is shown through numerical simulation that the proposed method can find the optimal allocation with less complexity compared with the exhaustive method.

Data processing can achieve desirable efficiency on a Graphics Process Unit cluster in the Compute Unified Device Architecture (CUDU) environment. However, the storage power and computing power of CUDA in the single-node environment has become the bottleneck of massive data processing. In order to process massive data efficiently in the CUDA environment, a computing framework for massive data processing is provided: CUDA on Hadoop. It combines CUDA and Hadoop that enhances the data throughput of CUDA applications by utilizing the distributed computing technology of Hadoop through a general interface for CUDA applications. In this paper, the details of the design and implementation of CUDA on Hadoop are illustrated as well.

Human tracking across disjoint camera views has obtained extensive concern with the increasing use of multi-camera surveillance with non-overlapping camera views. Human appearance matching is the key to continuous human tracking. Accurate human appearance matching among disjoint camera views is difficult due to various factors such as different view angles and illumination change. In this paper a method incorporating major color and spatio-texture features is proposed. Both features used are robust to illumination change. Major color describing global information is robust to the changes in view angles. A subtractive clustering method is used to extract major color representations from each object and then a formulation based on the normalized RGB color distance is proposed to weight the similarity among different major colors. The major color matching is calculated in an anniversary way. For objects owning similar color information, a spatio-texture feature concerning detailed information is exploited for further matching. In this paper, canny operators extracted from selected area are used as textural information. We test our approach indoor, and the results show that this method has successfully achieved a high matching accuracy in spite of general illumination change.

A fast approach is proposed for image dehazing using fuzzy system and hybrid evolutionary algorithm through fuzzy contrast enhancement. First, the RGB color space is converted into HSV color space and Gaussian membership function (MF) is used for the fuzzification. Then a parametric sigmoid function is used for the haze image contrast enhancement. Finally, an objective function combining with the entropy and the visual factors is optimal using a hybrid evolutionary algorithm (HEA). HEA is presented based on Partial Swarm Optimization (PSO) algorithm and Genetic algorithm (GA). On comparison, this approach is found applicable for image dehazing and better than the artificial ant colony system (AACS)-based method [

1

].

Cognitive code division multiple access (CCDMA) systems commonly use power control to reduce the interference between the users. But the power control is not always optimal. Probability control is a recently introduced method that allows better mitigation of multiple access interference in CCDMA networks. In this paper, we apply the probability control to CCDMA system. It was found that the performance reflected by the signal to noise plus interference ratio (SINR) resulting from probability control is better than the one resulting from power control by some simulation results.

In the stereo vision researches, global matching algorithms are widely used in the preference of high matching accuracy. This paper presents a new stereo matching algorithm based on global matching philosophy. The proposed method, which is called 3D dynamic programming-oriented matching method with filter-box acceleration, utilizes the edge information obtained by wavelet transform from the stereo images got from Middlebury University. We then compared the computational cost and matching accuracy among by the new method and other conventional stereo matching algorithms. We show that it gets advantages beyond those results. Finally, we make a conclusion about current works and look forward to feature jobs.

In this paper, we propose two fast dynamic descriptors Vertical-Time-Backward (VTB) and Vertical-Time-Forward (VTF) on spatial–temporal domain to catch the cues of essential facial movements. These dynamic descriptors are used in a two-step system to recognize human facial expression within image sequences. In the first step, the system classifies static images and then it identifies the whole sequence. After combining the visual-time context features with popular LBP, the system can efficiently recognize the expression in a single image, and is especially helpful in highly ambiguous ones. In the second step, we use the evaluation method through the weighted probabilities of all frames to predict the class of the whole sequence. The experiments were performed on 348 sequences from 95 subjects in Cohn–Kanade database and obtained good results as high as 97.6 % in seven-class recognition for frames and 95.7 % in six class for sequences.

Moving objects detecting and tracking is important for future Unmanned Aerial Vehicles (UAVs). We propose a new approach to detect and track moving objects from the flying UAV. First, estimate the global-motion of the background by tracking features selected by KLT algorithm from frame to frame. In order to avoid features located on the foreground objects participating in motion estimation, feature effectiveness evaluation is employed. Then compensate the background with the transform model computed by RANSAC. Define the undefined area before applying frame difference method to the compensated frame and the current frame. Then initialize the tracking module with information obtained from the detecting module, which overcomes shortcomings of artificial orientation of traditional tracking algorithms. For tracking fast and robustly from UAVs, we design a new tracking algorithm by fusing Kalman prediction and Mean Shift Search together. The experimental results presented effectiveness of the whole detecting and tracking approach.

Based on manifold ways of perception, this paper describes a novel method of object recognition and pose estimation within one integrated work. This method was inspired by bionic pattern recognition and manifold learning. Based on the principle of homology-continuity, we establish shortest neighborhood graph (SNG) for each class and regard it as a covering and triangulation for the hypersurface that the training data distributed on. For object recognition task, we propose a simple but effective classification method, named SNG-KNN. For pose estimation, local linear approximation method is adopted to build a local map between high-dimensional image space and low-dimensional manifold. The projective coordinates on manifold can depict the pose of object. Experiment results suggest that the recognition performance of our approach was similar and sometimes better compared to the SVM method; moreover, the pose of object can be estimated.

The recognition of traffic signs in natural environment is a challenging task in computer vision because of the influence of weather conditions, illuminations, locations, vandalism, and other factors. In this paper, we propose a vision-based traffic sign recognition system for the real utilization of intelligent vehicles. The proposed system consists of two phases: detection and recognition. In detection phase, we employ simple vector filter for chromatic/achromatic discrimination and color segmentation followed by shape analysis to roughly divide traffic signs into seven categories according to the color and shape properties. The Pseudo-Zernike moments features of the extracted candidate traffic sign regions are selected for recognition by random forests which combines bootstrap aggregating (bagging) algorithm and random feature selection to construct collections of decision trees and possesses excellent classification ability. The rationality and effectiveness of the proposed system is validated on our intelligent vehicle—Intelligent Pioneer from a great number of experiments.

Motor imagery-based BCI-FES rehabilitation system has been proved to be effective in the treatment of movement function recovery. Common Spatial Pattern (CSP) and Support Vector Machine (SVM) are commonly used in the feature extraction and classification of Two-classes motor imagery. However, motor imagery signals of stroke patients are irregular due to the damage of the specified brain area. Traditional CSP is not able to detect the optimal projection direction on such EEG data recorded from stroke patients under the interference of irregular patterns. In this paper, an adaptive CSP method is proposed to deal with these unknown irregular patterns. In the method, two models are trained and updated by using different subsets of the original data in every iteration procedure. The method is applied on the EEG datasets of several stroke subjects comparing with traditional CSP-SVM. The results also provide an evidence of the feasibility of our BCI-FES rehabilitation system.

Traffic data is the basis of Intelligent Transportation Systems (ITS). It is the key problem that how to fuse and share multi-resource heterogeneous data to provide comprehensive traffic information for ITS. This paper takes the Dynamic Traffic Routing System of Nanning City of China as example, multi-resource heterogeneous data fusion model is proposed, multi-resource heterogeneous database is generated, and then the GPS, Loop, and Video data are fused. The fusion results provide comprehensive traffic information for Dynamic Traffic Routing System and Traveler. Finally, the effectiveness of the proposed model was validated by the Dynamic Traffic Routing System of Nanning.

In the diagnostics, prognostics, and health management (DPHM) program for aircraft, enormous data, information, and knowledge relevant to the health states of aircraft are collected from various sources. The proper interpretation and using of these data and information constitute the basis for a sound decision making of maintenance activities. The raw data is pre-processed, extracted, combined, and integrated into reference information for decision making. Data fusion becomes a key technology at varied levels in this process This paper identifies and describes the role of data fusion in the context of modern DPHM program.

This paper addresses the problem of exact inference of probabilistic graphical models for multiscale segmentation of objects in the presence of dynamic backgrounds. Previous hidden Markov tree (HMT) based approaches have exploited the Expectation-Maximization (EM) algorithm to compute the optimal estimates of the multiscale parameters that maximize the likelihood function. However, the main problem with the EM algorithm is that it is a “

greedy

” method that converges to a local maxima on the log-likelihood surface. In this paper, we derive the Bethe free energy associated with the HMT which is a lower bound of cumulant energy function so as to recover multiscale posterior likelihoods exactly. This allows both inference and fusion of multiscale classification likelihoods to be computed through bottom-up likelihood estimation and up-bottom posterior inference of HMT. Experimental results on a frame of typical high-speed industrial inspection image demonstrate the correctness and robustness achieved by the proposed method.

Humanoid expression robot is a rising hot spot in the field of Artificial Intelligence (AI), and is very important to the human–computer harmonious interaction. People never stop the humanoid research of robots, which is the trend of the future. This paper wants to discuss three issues. First, can robot have cognitive emotion and show us? Second, how computer or analogous machines simulate complex emotions of human? Can the current emotion theory support and guide the development of cognitive robot? Third, combined with the current techniques and cognitive theory basis, how to design and build humanoid expression robot? Finally, the authors analyze the cognitive basis of emotion theory, and give the robot model and emotion model of cognitive expression.

We apply neural network to build up a detecting system of ink cells in gravure cylinder. First, ink cells images are gained in the images capturing device and histogram equalization. The edge of cells is extracted by use of Canny operator. We use different thresholds and experimental sigma values that compare to experimental results. Canny edge extraction operator is best when the value of sigma is 16. According to the image used in this research to determine the standard ink cells carving, the value of gaps

d

0

equals 125, the value of dark tone

s

0

equals 394, and so its standard value of gaps and dark tone are

d

0

± 10 and

s

0

± 10. The values of gravure outlets gaps and dark tone are measured, while

d

and

s

are in the scope of standard range, of which output 1 of the ink cells determined to pass and output 0 deemed to fail. Binarization images are obtained through adaptive threshold segmentation, which regards the value of gaps and dark tone as the characteristic value when they start to detect. Finally, we extract size and surface defects of ink cells for grading. Segmentation pictures are extracted by

K

-means clustering. The areas of ink cells are deemed to size characteristics. Then we classify the ink cells into two classes using neural network. The experimental results consider a neural network model that produces consequences.

Range-gated underwater laser imaging technology, which is very promising in oceanic research, deep sea exploration, and robotic works, is one of the most effective methods to suppress the effect of backward scattering of water medium. However, the special features of underwater laser images, such as speckle noise and nonuniform illumination, bring great difficulty for image segmentation. In this paper, an image segmentation algorithm which combines improved pulse coupled neural network with morphology is proposed. The morphology is applied to eliminate the speckle noise, while the cross-entropy is calculated as an optimization criterion for determination of the optimal segmentation. The experimental results of the proposed algorithm are compared with those of NCut, Mean-shift, Fuzzy C-means, and Watershed methods, and the quantitative evaluation confirms that the proposed algorithm is significantly superior to the other four algorithms in segmentation accuracy and robustness against speckle noise and nonuniform illumination.

In this paper, we introduce a new subspace learning algorithm in language recognition called locality preserving discriminant projection (LPDP). Total variability approach has been the state of art in language recognition, and it preserves most of the discriminant information of languages. Locality preserving projection (LPP) has been proved effective in language recognition, but it can only preserve the local structure of languages. LPDP method used in the total variability subspace can preserve both local structure and global discriminant information about the languages. Experiments are carried out on NIST 2011 Language Recognition Evaluation (LRE) database. The results indicate that LPDP language recognition system performs better than LPP language recognition system and total variability language recognition system in 30 s tasks. In addition, we also give the results of the total variability and LPDP language recognition systems on NIST 2007 LRE 30 s database.

Cognitive radio (CR) network can make an opportunistic access of spectrum licensed to a primary user (PU). The CR must perform spectrum sensing to detect active PU, thereby avoiding interfering with it. This chapter focuses on adaptively spectrum sensing, so that negative impacts to the performance of the CR network are minimized when CR users experience both stochastic data arrival and time-varying channel. Since the frequency of the spectrum sensing has a directly impact on system throughput and the probability of collision between the PU and CR user, the PU activity is modeled based on the characteristic analysis of the PU spectrum utilization. Based on that, an efficient adaptive sensing algorithm that takes into account the system stability, collision, and throughput is proposed. The CR users can make a balance between them by utilizing a “periodic control factor” which controls the adaptive adjustment of the spectrum sensing frequency. The simulation results indicate that the proposed algorithm has excellent performance on collision probability and throughput compared with conventional periodic spectrum sensing scheme. Meanwhile, it is shown that the proposed algorithm has low implementation complexity for practical applications.

Since Mars probe is very far from the earth during the flight, and the radio delay is up to tens of minutes, it is difficult to provide effective real-time information for the conventional ground-based radio navigation such as DSN of NASA. For probe in communication failure with the earth, it is very important to measure its position and speed based on its surrounding environment information, thus, probe autonomous navigation technology based on the target optical becomes a research focus. This paper first analyzes environment optical information easily perceived by the probe, then designs an autonomous position method based on optical information, at last, conducts a mathematical simulation to the positioning and speed measuring in the assumed Mars exploration Missions 2016. Simulation results show that the accuracy of the proposed method is very close to that of ground radio navigation, and the proposed method is feasible for engineering application.

A cognitive-heuristic framework for interconnecting analytical aerothermodynamics and mass-modeling parameters to heuristic optimizer is proposed. It evaluates a complex highly-integrated forebody-inlet configuration and representative hypersonic spaceplane based on minimal input data of flight altitude and Mach number only. SHWAMIDOF-FI design tool is used which incorporates salient features of multi-stage cognitive work approach integrated to heuristic optimization. Results show substantial improvement in geometric, performance, and flow parameters as compared to baseline configuration.

To satisfy the rapidly and accurately attitude tracking for spacecraft, artificial bee colony (ABC) algorithm is introduced to the controller parametric optimization of spacecraft attitude tracking. The spacecraft attitude tracking dynamics model, kinematics model and a sliding model controller using radial basis function neural network are build up. The concept of ABC algorithm is presented and the steps are also given. For the fitness function of ABC algorithm, the weighted index of error and angular velocity error with the simulation time are used. The optimization result compared between particle swarm optimization (PSO) and ABC algorithm shows the efficient of the ABC algorithm. The simulation result with the optimized controller shows that the controller could guarantee robustness against uncertainties and external disturbances increased.

LEO/MEO Multi-Layered Satellite Network (MLSN), consisting of low and medium earth orbit satellites, is capable of providing higher coverage and better service than most Single-Layered Satellite Network. Its performance, however, has been longly encumbered by obsolete routing protocols and algorithms. This paper takes the predictability of satellite movements into consideration, based on which a novel routing protocol—Predictable Satellite Network Routing Protocol (PSNRP), is proposed. In this protocol, all topology changes due to satellite movement are classified into predictable and unpredictable changes. This predictability assists to reduce the protocol overhead. The simulations show that except for obtaining better routing performance, PSNRP also successfully allocates calculation resources evenly among all nodes, separates user data from protocol control data, and achieves stronger robustness on undergoing satellite failures and link congestions.

This paper presents a more robust iterative closest point (ICP) approach for 2D point set registration. An inequality constraint of the rotation angle is introduced into the least square registration model which is solved by an extended ICP algorithm. At each iterative step of the algorithm, a closed-form solution for the rotation is obtained according to the monotonicity of the model with respect to the rotation angle. The proposed approach extends the convergence domain of the ICP algorithm, and it can be used much more widely. A series of 2D point set experiments on part B of MPEG-7 CE-shape-1 dataset prove that the proposed method is much more robust than ICP without increasing the computational complexity.

Obstacle detection is very important for Advanced Driving Assistance Systems (ADAS) and Unmanned Ground Vehicles (UGV) in on-road scene. The motion of vehicles is restricted by a pair of lane markings. This constraint makes obstacle detection in the on-road scene difference to the one in other robot application. Objects between lane markings are the most emergent obstacles and should be taken more attention. In this paper, we present an effective and efficient method to fuse lane marking information obtained from camera with LIDAR data to discriminate obstacles as within lane markings and outside lane markings.

A computationally efficient method for the problem of estimating the parameters—the central direction of arrival (DOA) and angular spread of coherently distributed source is presented. The proposed method is based on Schur–Hadamard product which enables the estimation of the central DOA decoupled from that of angular spread of the source. So an underlying rotational invariance structure is exploited. Then the key idea is to apply the propagator method to estimate the central DOA, which only requires linear operations. The angular spread is estimated by a proposed second-order statistics sequently, from which the closed solution of angular spread is derived. An advantage of this method over the classical subspace-based algorithm, such as ESPRIT and MUSIC for distributed source, is that it does not apply any searching. Numerical examples illustrate the performance of the method.

In many applications, the received image is degraded by unknown blur and noise. Traditional blind image deconvolution algorithms have drawback of noise amplification. For robustness against the influence of noise, this paper proposes a novel blind image deconvolution algorithm by combining the support vector regression (SVR) approach and the total variation approach. The proposed algorithm has a lower computational complexity and a good performance in image denoising and image deblurring. Illustrative examples show that the proposed blind image deconvolution algorithm and has better performance in improvement signal-to-noise ratio than two traditional blind image restoration algorithms.

This study investigated the interaction of music emotional styles and tempo in behavior and eye movement in 3D driving simulator. The main results indicated that music emotional styles affected driving behavior and eye movement in both slow tempo and fast tempo conditions, but the means of emotional styles affecting driving behavior and eye movement differed on tempo. There was significant emotional styles difference on the number of mistakes and mean fixation duration in fast tempo condition, while there was emotional styles effect on time perception and vertical spread of visual search in slow tempo condition. According to these results, we suggested that drivers should choose peaceful and cheerful music while driving.

Despite being widely adopted and rigorously followed in many successful pedestrian detectors, the original HOG (Histogram of Oriented Gradients) descriptor are in fact NOT optimally tuned for pedestrian detection. To address this issue, we quantitatively investigate the interplay among different HOG parameters, in particular that among the cell size, aspect ratio, and detection window size, which makes it possible to jointly tune these parameters to achieve better pedestrian detection performance. In addition, we extend the training procedure of the original HOG-based detector of Dalal et al. through presenting an automatic positive sample generation algorithm, introducing LSVM (Latent SVM) to iteratively optimize the positive training samples, and adopting a hard negative mining method. To verify the effectiveness of our improved detector, we conduct extensive experiments on INRIA Person, TUD-Brussels and Caltech Pedestrians datasets. On all these datasets, our detector outperforms significantly the original HOG detector of Dalal et al.

This paper introduces a reconfigurable array synthesis method using the correlation weightings of Smooth Local Trigonometric Base (SLTB) for line antenna array and investigates its beam pattern characteristics. The beam pattern can be reconfigured by adjusting the overlapping coefficiency

$$ r $$

of the bell-shaped function, the number

K

and the frequency spacing

$$ \Updelta f $$

of the SLTB. A fast and convenient reconfigurable array synthesis algorithm is proposed according to the evolvement rule of the beam pattern. The proposed method provides a nearly optimum first sidelobe level and gradually decaying sidelobes compared with Chebyshev weighting. Moreover, its computational complexity is

O

(

N∙K

2

) while the one of Chebyshev is

O

(

N

3

).

Brain–computer interface (BCI) is a new way for man–machine interaction with wide applications, in which steady-state visual evoked potentials (SSVEP) is a promising option. However, many characteristics of SSVEP show great user variation. So parameter optimization and channel selection for each subject were applied to improve the performance of BCI. These optimizations limit the practical applicability of the SSVEP-based BCI. The use of a canonical correlation analysis (CCA) method for multi-channel SSVEP in the time domain detection showed highly increased detection accuracy, but it is sensitive to the noise when the stimulate frequency is low. In this paper, a method of CCA in the frequency domain is presented for classifying multi-channel SSVEPs. First overlapping average is conducted on the original training signals. Then fast Fourier transform (FFT) is used to transform the signals from time domain to frequency domain to produce the reference data. Finally, according to the correlation coefficients of the new data and the references in the frequency domain, the SSVEP is classified. The experimental results show the enhanced accuracy of our method when applied to low stimulate frequencies.

Control of humanoid robot behavior with the mind begins a new era of robotics research. One of the critical issues in this research is how to acquire the brain signals with high quality which are correlated to humanoid robot behavior. In order to improve subjects’ concentration on their mental activities during tests, we develop a stimuli module in the Cerebot system, consisting of a Cerebus neural data acquisition system and a Kumotek robot with 20 degrees of freedom or a NAO robot with 25 degrees of freedom. We present the experimental procedures for acquiring brain signals of imagining humanoid robot walking behavior by using movies of robot walking or real robot walking activities. We record two groups of brain signals correlated to mental activities of six robot walking behavior. Finally, we present a demonstration of controlling the humanoid robot walking behavior using the phase coding mechanisms of the

Delta

rhythms.

Category-independent object detection is extremely useful for many robot vision tasks. Most existing methods rank a lot of regions by measuring their object-likeness. However, to obtain a sufficient object covering rate too many regions need to be sampled. In this paper, we present a novel method that directly detects and localizes category-independent objects. We develop a novel model which is named as “mixed robust higher-order conditional random field” model which combines 2D and 3D data into a uniform framework. A set of novel features is developed based on 2D and 3D saliency and oversegments. The potentials used in this model are computed from these features. Extensive experiments are carried out on a public RGB-D dataset. By comparison with state-of-the-art ranking methods, the experimental results show the comparable performance of category-independent object detection without sampling a large number of extra regions.

The orbit attribution is the most important information on space object. Therefore, parallel management is channeled when complicated and systematic projects are decided and evaluated. Its method and process are applied to the study of system simulation. Traditionally, large-scale simulation and control of the system depended on single model simulation and traditional control approaches for making the target system under control at the only past time. However, the parallel management methods’ application will get the system under control and be accurate for a long time in the future. The paper briefly reviews ACP theory and data-based iterative learning control, then points out the necessity and feasibility of the combination of parallel control and iterative control by putting out the structure of NN iterative learning controller. The space objects prove the methods by calculating the orbit attribute at the end of the paper. This paper focuses on parallel management principles and makes further studies of its implementation methods for trying to offer some theory achievements and practical experiences for future researches.

Joint probability density algorithm (JPDA) is extended to multi-sensor multi-target detection based on ‘Clean’ method. Key content and characteristic of JPDA in false intersection points eliminating are introduced. An iterative method for multi-sensor multi-target detection is put forward. Based on target that has been detected by peak searching, an inverse probability density function is proposed which is able to eliminate both the detected target and false intersection points. A new joint probability density matrix is constructed and all the targets can be detected through iterative process. The feasibility and validity are verified through simulation.

In this paper, we provide a solution of teaching a robot to perform tasks through human demonstration. On the one hand, we consider many everyday complex task manipulations made up of manipulation primitives, consisting of a series of sequential rotations and translations. On the other hand, we design demonstration primitives which decompose a demonstrated task. We use Kinect sensor to locate some hot points’ position of a teacher’s hand during demonstration, in order to gain the axes of those rotations and translations which will support the manipulation primitives that will contribute to building task descriptors. Based on this, we quote a descriptor to represent this manipulation primitive. We also teach the robot where to start and where to end once a manipulation primitive is operating via recording the special two points’ coordinates in a real scene while demonstration, which could reinforce robots’ learning to those effective and sequential demonstrations and shorten the learning time. One manipulation primitive corresponds with one demonstration primitive. Two kinds of primitives will be connected to the axes of trajectory. After that, we perform experiments to discuss the universality of this framework.

An innovative universal automatic driving system for intelligent vehicles is designed in this research. With a few mechanical modifications of four subsystems, i.e., steering system, breaking system, throttle system, and gearing system of the original manned vehicles, and the installation of an automatic control device, this system is finally established by connecting each subsystem and upper computer via CAN bus. It is proved by vehicle tests that upper computer can precisely control the underlying subsystem through automatic driving system, and this system can be used not only for vehicle bench test, but also for automatic driving of intelligent vehicles, which establishes a foundation for further researches of unmanned and intelligent control of vehicles.

Auditory steady-state response (ASSR) is a kind of signal which is phase-locked to the stimulus onset. Consequently, the methods of detecting ASSR are generally based on the feature of phase-locking. In this paper, ASSR in 40 and 90 Hz were recorded from 13 subjects. The phase values were calculated while the subjects attended to the tones. And it was found that besides S7, for each frequency, the phase of ASSR were fairly constant across all 12 subjects. For examples, the phase of ASSR responding to 40 Hz was mainly distributed around 3/4pi. Based on the results, the method of

V

test was introduced in this paper. By comparing the results obtained by

V

test and Rayleigh test, we found that

V

test is a more effective method for detecting ASSR.

In this paper, a novel neural network (NN)-based adaptive dynamic surface control (DSC) is proposed for inverted pendulum system. This scheme overcomes the problem of “explosion of complexity” which is inherent in the traditional backstepping technique. Meanwhile, the effect of input saturation constrains is considered in the control design. All the signals in the closed-loop system are proved uniformly ultimately bounded. Finally, the experimental platform simulation results are used to demonstrate the effectiveness of the proposed scheme.

This paper describes an approach of using image processing and patters classification techniques for navigating the unmanned aircraft vehicle in known irregular environment. In the case of 2D path planning, a feasible flight path connecting the start and goal point can be regarded as a separating surface that divides the space into two regions. This suggests a dual problem of first dividing the whole space into such two regions and then picking up the boundary as a path. We use support vector machine to solve this dual problem. SVM can generate a nonlinear separating surface based on the margin maximization principle. First, we generate a novel search space which contains flyable and no-fly regions from 3D surface of minimum risk and pick up key obstacle points as samples. Second, a safe and smooth path is generated through SVM. Results from simulations show that the path planner is able to plan an optimal path efficiently due to the simplicity of the search space.

It is widely recognized that one of the most effective ways to solve urban traffic problems is developing public transport system, especially urban rail transit system. The estimation of passenger flow distribution, as an important part of travel demand analysis of urban rail transit, is the prerequisite of the operation organization and management of urban rail transit system, especially when a new line is put into operation. This paper proposes a new passenger flow distribution model, which is based on disaggregate model approach and conforms the aggregated historical passenger flow data to disaggregate data through representative individual method. Influencing factors including travel time, attracted traffic flow, land-use type, intensity around station, and so on are considered. Using the historical passenger flow data of urban railway system in Beijing before and after Line 4 is put into operation, the model is built and the estimation accuracy evaluated. The result shows that the disaggregate model is more accurate than the conventional aggregate single restraint gravitational model.

This paper deals with the Information consensus problem in distributed systems under local communication and dynamically switching interaction topologies. We show conditions under which consensus is reached under switching directed information exchange topologies. We propose distributed model predictive control schemes that take into account constraints on the agents input and show that they guarantee consensus under mild assumptions. Simulation results show that the proposed scheme is effective under both fixed and dynamically switching interaction topologies.

To examine the effect of different visual information on path integration, we built four virtual environments with purecolor, texture, landmarks, and both texture and landmarks as visual cues, and used a point-to-origin task to estimate people’s performance on path integration. Though constant bias existed under all visual environments, people were found to use different strategies according to different visual environments. To point back to the origin of the path, visual path integration was more likely to be achieved in the environments containing landmarks or both texture and landmarks, while in the environments with only texture or pure-colors, people turned to rely on temporal duration and the layout of the path. It was suggested that landmarks were more useful visual cues in virtual navigation.

This paper retrospects the domestic and overseas development process of the fault diagnosis technology. Combined with today’s cruise missile with turbofan engine production, testing, and use of the actual characteristics of fault diagnosis technology research and analysis in the field range of applications, the value and significance of the shells with the turbofan engine fault diagnosis technology are introduced. In addition, this paper focused on the instance of the class fault diagnosis system study. And the development trends of future missile turbofan engine fault diagnosis technologies are summarized.

To improve the capturing speed of a GPS software receiver, a novel acquisition scheme is presented. First, by compensating the frequency of the intermediate frequency signal to improve its cyclicity, the signal-to-noise ratio (SNR) of the intermediate frequency signal can be improved by using coherent average algorithm. Second, the C/A code start of every satellite is searched through the delay−multiply configuration, then the delay and accumulation unit is put forward to pretreat the multi-frequency signals. Lastly, the separation and estimation of Doppler-shift components are fulfilled by frequency domain analysis on post-treatment signals; therefore, the two-dimensional searching process in conventional acquisition scheme is transformed into two one-dimensional searching processes in the proposed acquisition scheme. In the proposed acquisition scheme, a shorted fast Fourier transform (FFT) version is adopted, which can greatly reduce the computation burden without degradation on performance. Sampled data from GPS receiver and simulated GPS signals are used for simulation experiments. Simulation results indicate that the proposed acquisition scheme is an effective one by much decreasing the acquisition time.

The process of different human manipulating a specific object in hand obeys very similar operating steps. The hand movement can be modeled and generalized into action gist to guide other human or robots to execute the specific in-hand manipulation task. This paper suggests a kind of action gist similar to the way humans learn to represent the five finger hand motions in in-hand manipulation. Our method is based on Gaussian Markov Random Field that processes data-glove values to obtain the action gist. Several experiments are carried out to discuss the performance of the proposed methods.

This publication describes an innovative intelligent omnivision video system, that stitches the images of four cameras together, resulting in one seamless image. This way the system generates a 360° view of the scene. An additional automatically controlled pan-tilt-zoom-camera provides a high resolution view of user defined regions of interest (ROI). In addition to the fusion of multiple camera images, the system has intelligent features like object detection and region-of-interest detection. The software architecture features configurable pipelines of image processing functions. All different steps in the pipeline like decoding, feature extraction, encoding, and visualization are implemented as modules inside this pipeline. It is easily possible to rearrange the pipeline and add new functions to the overall system. The pan-tilt-zoom camera is controlled by an embedded system that has been developed for this system. GPU-accelerated processing elements allows real-time panorama stitching. We show the application of our system in the field of maritime surveillance, but the system can also be used for robots.

The approach presented shows possible ways of improving scene analysis to achieve more reliable and accurate object recognition in the context of mobile robotics. The centralized architecture combines different feature detectors with active modalities, such as change of perspective or influencing the scene. It opens possibilities for the use of 2D detectors and extends the results to 3D. In combination with mixed reality, it offers the possibility of evaluation of the developed system as well as increased efficiency. The architecture developed and the preliminary results are presented. The work goes a step in the direction of active intelligent perception.

The development of humanoid robots for helping humans as well as for understanding the human cognitive system is of significant interest in science and technology. How to bridge the large gap between the needs of a natural human-robot interaction and the capabilities of recent humanoid platforms is an important but open question. In this paper we describe a system to teach a robot, based on a dialogue in natural language about its real environment in real time. For this, we integrate a fast object recognition method for the NAO humanoid robot and a hybrid ensemble learning mechanism. With a qualitative analysis we show the effectiveness of our system.

The Verbally Assisting Virtual-Environment Tactile Maps (VAVETaM) approach proposes to increase the effectiveness of tactile maps by realizing an intelligent multi-modal tactile map system that generates assisting utterances that generates assisting utterances acquiring survey knowledge from virtual tactile maps. Two experiments in German conducted with blindfolded sighted people and with blind and visually impaired people show that both types of participants benefit from verbal assistance. In this paper, we report an experiment testing the adaptation of the German prototype to be useable by Chinese native speakers. This study shows that the VAVETaM system can be adapted to Chinese language with comparable small effort. The Chinese participants’ achievement in acquiring survey knowledge is comparable to those of the participants in the German study. This supports the view that human processing of representationally multi-modal information is comparable between different cultures and languages.

Image denoising is a traditional problem which has been tackled using a variety of conceptual frameworks and computational tools. Total variation-based methods have proven to be efficacious toward solving image noise removal problems. Its purpose is to remove unnecessary detail and achieve optimal performance in terms of mean squared error (MSE), a metric that has been widely criticized in the literature due to its poor performance as an image visual quality assessment. In this work, we use structural similarity (SSIM) index, a more accurate perceptual image measure, by incorporating it into the total variation framework. Specifically, the proposed optimization problem solves the problem of minimizing the total gradient norm of restored image and at the same time maximizing the SSIM index value between input and reconstructed images. Furthermore, a gradient descent algorithm is developed to solve this unconstrained minimization problem and attain SSIM-optimal reconstructed images. The image denoising experiment results clearly demonstrate that the proposed SSIM-optimal total variation algorithm achieves better SSIM performance and better perceptual quality than the corresponding MSE-optimal method.

Learning from sensor data plays an important role in the field of robotic research, especially in dexterous robotic hand grasping. The manuscript puts efforts on learning from tactile dynamic process during robotic hand grasping. A piecewise linear dynamic systems and a group of models are presented, under the guidance of which, proper gesture according to different types of targets could then be selected to facilitate stable and accurate grasping. This is evaluated on the experimental testbed and shows promising results.

We consider the problem of traffic sign recognition in image sequences. In many cases, image sequences of traffic signs can be collected from consecutive videos and these images have high correlation with each other. While traditional traffic sign recognition approaches focus on how to extract better features and design more powerful classifiers, most of these methods neglected this correlation. In this paper, we introduce the low-rank matrix recovery model to exploit the correlation among images with similar appearances to enhance feature representation. By recovering the underlying low-rank matrix from the original feature matrix consists of feature vectors of image sequences, we are able to attenuate the influence of corruption, such as noise and motion blur. Experiments are conducted on GTSRB dataset to evaluate our method, and noticeable performance gain is observed by using low-rank matrix recovered from original matrix. We obtain very impressive results on several super-class accuracy while get comparable performance with state-of-the-art results on global accuracy.

We propose a linear spatial pyramid matching using locality-constraint linear coding for SAR image classification based on MSTAR database. Recently, works have little consideration about targets’ randomly distributed poses when applying sparse coding in coding scheme. We do the preprocessing of pose estimation to generate over-complete codebook and therefore reduce reconstruction error. SIFT descriptors extracted from images are projected into its local-coordinate system by Locality-constrained linear coding instead of sparse coding. Locality constraint ensures similar patches will share similar codes. The codes are then pooled within each sub-region partitioned according to spatial pyramid and concatenated to form the final feature vectors. We use max-pooling which is more salient and robust to local translation. With linear SVM classifier, the proposed approach achieves better performance than traditional ScSPM method.

Electroencephalography is a commonly used tool for presurgical evaluation of epilepsy patients. In this paper, we present a quantitative evaluation of interictal high frequency oscillations (HFOs) in scalp Electroencephalographies (EEGs) for epileptogenic region localization. We process multichannel EEGs using time–frequency spectral analysis in order to detect HFOs in each EEG channel. Comparison between the results of time–frequency analysis and visual assessment is performed to verify the reliability of time–frequency analysis. Later,

$$t$$

-test and Pearson correlation analysis are performed to analyze the relationships between ictal HFOs and interictal HFOs. The high correlations between interictal and ictal HFOs imply that interictal HFOs, like ictal HFOs, are valuable in localizing the epileptogenic region. As a result, scalp interictal HFOs are valuable in epileptogenic region localization for presurgical evaluation of epilepsy patients. It holds great potential for reducing the long delay before patients can be referred for surgery.