Skip to main content
main-content

Über dieses Buch

Computer Vision and Robotic is one of the most challenging areas of 21st century. Its application ranges from Agriculture to Medicine, Household applications to Humanoid, Deep-sea-application to Space application, and Industry applications to Man-less-plant. Today’s technologies demand to produce intelligent machine, which are enabling applications in various domains and services. Robotics is one such area which encompasses number of technology in it and its application is widespread. Computational vision or Machine vision is one of the most challenging tools for the robot to make it intelligent. This volume covers chapters from various areas of Computational Vision such as Image and Video Coding and Analysis, Image Watermarking, Noise Reduction and Cancellation, Block Matching and Motion Estimation, Tracking of Deformable Object using Steerable Pyramid Wavelet Transformation, Medical Image Fusion, CT and MRI Image Fusion based on Stationary Wavelet Transform. The book also covers articles from applications of soft computing techniques such as Target Searching and Tracking using Particle Swarm Optimization, PSO-based Functional Artificial Neural Network, etc. The book also covers article from the areas of Robotics such as Solar Power Robot Vehicle, Multi Robot Area Exploration, Intelligent Driving System based on Video Sequencing, Emotion Recognition using MLP Network, Identifying the Unstructured Environment.

Inhaltsverzeichnis

Frontmatter

Tracking of Deformable Object in Complex Video Using Steerable Pyramid Wavelet Transform

Abstract
Tracking of deformable moving object in complex video is one of the challenging problems in computer vision. Random motion, varying size and shape of the object, and varying background and lighting conditions make the tracking problem difficult. Many researchers have tried to handle this problem using spatial domain-based methods, but those methods are not able to handle movement of object properly in case of varying size, shape, and background of object. In this paper, we have proposed a tracking algorithm for deformable video object. The proposed method is based on the computation of steerable pyramid wavelet transform. Coefficients at different level of decomposition and velocity of object are used to predict object location in consecutive frames of video. The approximate shift-invariance and self-reversibility properties of steerable pyramid wavelet transform are useful for tracking of object in wavelet domain. The translation in object is well handled by shift-invariance property, while self-reversibility property yields to make it useful to handle object boundaries. Experimental results of the proposed method and its comparison with other state-of-the-art methods show the improved performance of the proposed method.
Om Prakash, Manish Khare, Rajneesh Kumar Srivastava, Ashish Khare

Use of Different Features for Emotion Recognition Using MLP Network

Abstract
Emotion recognition of human being is one of the major challenges in modern complicated world of political and criminal scenario. In this paper, an attempt is taken to recognise two classes of speech emotions as high arousal like angry and surprise and low arousal like sad and bore. Linear prediction coefficients (LPC), linear prediction cepstral coefficient (LPCC), Mel frequency cepstral coefficient (MFCC) and perceptual linear prediction (PLP) features are used for emotion recognition using multilayer perception (MLP).Various emotional speech features are extracted from audio channel using above-mentioned features to be used in training and testing. Two hundred utterances from ten subjects were collected based on four emotion categories. One hundred and seventy-five and twenty-five utterances have been considered for training and testing purpose.
H. K. Palo, Mihir Narayana Mohanty, Mahesh Chandra

Rectangular Array Languages Generated by a Petri Net

Abstract
Two different models of Petri net structure to generate rectangular arrays have already been defined. In array token Petri net structure, a transition labeled with catenation rule is enabled to fire only when all the input places of the transition have the same array as token. In Column row catenation Petri net structure, the firing rules differ. A transition labeled with catenation rule is enabled to fire even when different input places of the transition contain different arrays. The firing rule associated with a transition varies in the two models. Comparisons are made between the generative capacity of the two models.
Lalitha D.

Medical Image Fusion Using Local Energy in Nonsubsampled Contourlet Transform Domain

Abstract
Image fusion is an emerging area of image processing. It integrates complementary information of different source images into a single fused image. In the proposed work, we have used nonsubsampled contourlet transform for fusion of images which is a shift-invariant version of contourlet transform. Along with this property, it has many advantages like removal of pseudo-Gibbs phenomenon, better frequency selectivity, improved temporal stability, and consistency. These properties make it suitable for fusion application. For fusing images, we have used local energy-based fusion rule. This rule depends on the current as well as the neighboring coefficients. Hence, it performs better than single coefficient-based fusion rules. The performance of the proposed method is compared visually and quantitatively with contourlet transform, curvelet transform, dual-tree complex wavelet transform, and Daubechies complex wavelet transform-based fusion methods. To evaluate the methods quantitatively, we have used mutual information, edge strength, and fusion factor quality measurements. The experimental results show that the proposed method performs better and is more effective than other methods.
Richa Srivastava, Ashish Khare

Machine Learning and Its Application in Software Fault Prediction with Similarity Measures

Abstract
Nowadays, the challenge is how to exactly understand and apply various techniques to discover fault from the software module. Machine learning is the process of automatically discovering useful information in knowledgebase. It also provides capabilities to predict the outcome of future solutions. Case-based reasoning is a tool or method to predict error level with respect to LOC and development time in software module. This paper presents some new ideas about process and product metrics to improve software quality prediction. At the outset, it deals with the possibilities of using lines of code and development time from any language may be compared and be used as a uniform metric. The system predicts the error level with respect to LOC and development time, and both are responsible for checking the developer’s ability or efficiency of the developer. Prediction is based on the analogy. We have used different similarity measures to find the best method that increases the correctness. The present work is also credited through introduction of some new terms such as coefficient of efficiency, i.e., developer’s ability and normalizer. In order to obtain the result, we have used indigenous tool.
Ekbal Rashid, Srikanta Patnaik, Arshad Usmani

Various Strategies and Technical Aspects of Data Mining: A Theoretical Approach

Abstract
In this paper, we are going to look at a very interesting aspect of database management, namely data mining and knowledge discovery. This area is attracting interest from not only researchers but also from the commercial world. The utility of data mining in commerce is more interesting than perhaps research areas. This has also raised many debates such as rights of privacy, legality and ethics, and rights to non-disclosure of information. It has someway opened a Pandora’s box. Only, time will tell whether it is on the whole destructive or constructive; nonetheless, technology is not as such absolutely constructive or destructive; it only depends on how it is brought into use. In this paper, we have discussed about the technical aspects of data mining and what are the different strategies of data mining. Its sections give many technical aspects for various data mining and knowledge discovery methods, and we have given a rich array of examples and some are drawn from real-life applications.
Ekbal Rashid, Srikanta Patnaik, Arshad Usmani

CBDF-Based Target Searching and Tracking Using Particle Swarm Optimization

Abstract
Target tracking and searching are the very important problems in robotics. It can be used in many variations like path planning where the objective is to reach up to target without collide with obstacle, or it may be used in the places where some source is needed to find. Here nature inspired PSO algorithm is used to solve this problem with the help of multi robot system. Use of multi robots will find the target fast. Here new concept clustering based distributing factors (CBDF) is introduced to scatter the robots in environment to search and track the target. This CBDF method divides the area into regions. Different robots use coordination to reach up to these targets. For the movement of robots, PSO is used because it can be considered as minimization problem with respect to minimize the path length. Here, four parameters move, time, coverage, and energy are calculated to reach up to target. At last, we are showing the results for both known and unknown target problem.
Sanjeev Sharma, Chiranjib Sur, Anupam Shukla, Ritu Tiwari

Real-Time Noise Canceller Using Modified Sigmoid Function RLS Algorithm

Abstract
In this paper, modified sigmoid function RLS (MSRLS) algorithm is proposed for online noise cancellation from audio signals. The experiments are performed using TMS320C6713 processor with code composer studio (CCS) v3.1. The performance of RLS and MSRLS algorithms is evaluated and compared for noisy signals with car noise, F16 noise, and babble noise at −5, 0, and 5 dB SNR levels. The proposed MSRLS algorithm has shown a maximum of 2.03 dB improvement in SNR over RLS algorithm at input signal of −5 dB SNR with F16 noise. The proposed MSRLS algorithm has also shown decrement in mean square error (MSE) at all SNR levels for all noises in comparison with RLS algorithm.
V. K. Gupta, D. K. Gupta, Mahesh Chandra

Multiscale Local Binary Patterns for Facial Expression-Based Human Emotion Recognition

Abstract
Facial expression is an important cue for emotion recognition in human behavior analysis. In this work, we have improved the recognition accuracy of facial expression recognition and presented a system framework. The framework consists of three modules: image processing, facial features extraction, and facial expression recognition. The face preprocessing component is implemented by cropping the facial area from images. The detected face is downsampled by bilinear interpolation to reduce the feature extraction area and to enhance execution time. For extraction of local motion-based facial features, we have used rotation-invariant uniform local binary patterns (LBP). A hierarchical multiscale approach has been adopted for computation of LBP. The selected features were fed into a well-designed tree-based multiclass SVM classifier with one-versus-all architecture. The system is trained and tested with benchmark dataset from JAFFE facial expression database. The experimental results of the proposed techniques are effective toward facial expression recognition and outperform other methods.
Swati Nigam, Ashish Khare

Development of a New Algorithm Based on SVD for Image Watermarking

Abstract
The research on watermarking has been increasing day-by-day since past decade. It has been largely driven by its important applications in digital copyrights management and protection. To provide more watermarks and to minimize the distortion of the watermarked image, a novel technique is presented in this paper. In this paper, the singular value decomposition (SVD)-based image watermarking scheme is proposed. The output result of SVD is more secure and robust. SVD is often used to develop robust watermarking algorithms. However, the SVD-based algorithms exhibit false-positive problem and pose security concern. In this work, we try to overcome this problem. In the proposed schemes, the host image is first decomposed into sub-bands by applying discrete wavelet transform (DWT). The watermark image is embedded in all the sub-bands by modifying the singular values of each sub-band. Next to it, we propose to encrypt and embed the singular values of the watermark image instead of original singular values. RSA algorithm has been used for the encryption process. Peak signal-to-noise ratio (PSNR) is used to measure the imperceptibility of the proposed schemes. The simulation result shows its efficacy.
Arun Kumar Ray, Sabyasachi Padhihary, Prasanta Kumar Patra, Mihir Narayan Mohanty

Nonlinear System Identification Using Clonal Particle Swarm Optimization-based Functional Link Artificial Neural Network

Abstract
In this paper, the clonal particle swarm optimization (C-PSO)-based functional link artificial neural network model (FLANN) has been applied for the identification of a nonlinear system. System identification in different challenging situations such as noisy and time varying environments has been a matter of great concern for researchers and scientists for the last few decades. Other variants of the FLANN network, trained with some of the optimization techniques, such as the genetic algorithm (GA), particle swarm optimization (PSO), and comprehensive learning particle swarm optimization (CLPSO) have also been applied in this interesting field of research. The proposed C-PSO algorithm is based on the clonal principle in a natural immune system. In the C-PSO, the essence of the clonal operator is to generate a set of clone particles near the expected candidate solution. Hence, the search spaces are enlarged, and the diversity of the cloned particles is increased to avoid trapping in local minima. The simulation study reveals the superiority of the proposed C-PSO-based FLANN model, in terms of the convergence rate, over other competitive networks. The performance comparison is also carried out based on the computational complexity.
Kumar Gaurav, Sudhansu Kumar Mishra

A Steganography Scheme Using SVD Compressed Cover Image

Abstract
Steganography is one of the popular tools for secure transmission of confidential data through public channels like the Internet. The common approach of steganography technique is to use any meaningful uncompressed multimedia like image or video as a carrier. The Internet comprises limited channel bandwidth so multimedia data like image or video are considered as a compressed form before their transmission through Internet. So in this paper, we have opted compressed image as a cover media for embedding the secret message and the modified compressed image components will preserve the visual content when it will be reconstructed. Singular value decomposition was employed in cover image for compression. In order to enhance the security, the secret message was embedded into compressed image components where the coefficients were not selected in sequence order. So in the proposed work, a straightforward message extraction process will not be applicable. We have tested the proposed scheme on some standard test images and satisfactory results were achieved.
Kshiramani Naik, Shiv Prasad, Arup Kumar Pal

Particle Swarm Optimization-based Functional Link Artificial Neural Network for Medical Image Denoising

Abstract
In this paper, a new computationally fast and efficient adaptive digital image filter has been proposed for denoising of digital medical image corrupted with additive white Gaussian noise. A particle swarm optimization-based functional link artificial neural network (FLANN) has been applied for this interesting and challenging problem. The three others competitive networks based on artificial neural network such as multilayer perceptron (MLP), direct linear artificial feed-through neural network (DLFANN), and LMS-based FLANN have also been applied for this purpose. The quantitative analysis of the proposed algorithm has been carried out by taking the peak signal to noise ratio (PSNR) and mean square error (MSE) as two parameters. Experimental results demonstrate the efficiency and effectiveness of the proposed algorithm.
Manish Kumar, Sudhansu Kumar Mishra

Multi-robot Area Exploration Using Particle Swarm Optimization with the Help of CBDF-based Robot Scattering

Abstract
Robot area exploration is a very important task in robotics because it has many applications in real-life problem. So, this is always a very interesting field for researches. This paper presents a new method for multi-robot area exploration. Here, first the environment is divided into partition. In each partition, the robot is deployed randomly. Each partition is explored separately by robot. For the movement of robots, well-known particle swarm optimization algorithm is used. Here mainly concentrated on the multi-robot-coordinated exploration for unknown search spaces where decisions made by bio-inspired algorithms for movement and thus helping in exploration. For better and fast exploration, robot should be scattered in different directions; for this purpose, new clustering-based distribution method is used. The proposed method is tested on different simulated environments that are considered as indoor and outdoor environments. Different parameters such as move, coverage, energy, and time are calculated. The results show that method works well for both environments.
Sanjeev Sharma, Chiranjib Sur, Anupam Shukla, Ritu Tiwari

R–D Optimization of EED-based Compressed Image Using Linear Regression

Abstract
Image compression technique minimizes the size in bytes of a graphics file without degrading the quality of the image to an unacceptable visual level. The performance is driven by the trade-off between rate (R), distortion (D), and complexity of the coding algorithm. This piece of work deals with the simple yet effective method of image-dependent R–D optimization by optimizing the cost function of the constrained Lagrangian multiplier. A simple linear regression is employed to estimate the cost and the slope of the convex haul in the R–D plane. The image is compressed by SPIHT algorithm after it was subjected to anisotropic diffusion to provide a powerful zero tree pyramid. The Perona–Malik diffusion preserves the edge and boundary information by smoothening the whole image. Extensive simulation for the proposed diffusion platform provides better results in terms of PSNR and visual quality, and the estimation approximates the actual values with minimum residuals.
Bibhuprasad Mohanty, Madhusmita Sahoo, Sudipta Mohapatra

Comparative Analysis of Image Deblurring with Radial Basis Functions in Artificial Neural Network

Abstract
Radial basis functions are used in many fields of mathematics and image analysis. In this paper, we have used linear RBF, cubic RBF, multi-quadratic RBF, inverse multi-quadratic RBF and Gaussian RBF for the reconstruction of blurred images. Simulations and mathematical comparisons show that Gaussian RBF gives better result with respect to the other RBF methods for images reconstruction in artificial neural network.
Abhisek Paul, Paritosh Bhattacharya, Prantik Biswas, Santi Prasad Maity

A Hybrid Approach for Iris Recognition in Unconstrained Environment

Abstract
Iris recognition is one of the emerging areas as the demand for security in social and personal areas is increasing day by day. The most challenging step in the process of iris recognition is accurate iris localization. As it significantly affect the further processing of feature extraction and template matching stages. Traditional algorithms work efficiently and accurately in localizing iris from eye images taken in constrained environmental conditions. But their accuracy gets affected when eye images are taken in unconstrained environmental conditions. The proposed algorithm starts with determining the exact location of iris even in the presence of specular highlights and non-ideal environmental conditions. It also works well for noisy iris images, and the accuracy of segmentation stage has increased when intuitionistic fuzzy is applied on the UBIRIS v2 iris images.
Navjot Kaur, Mamta Juneja

Fractal Video Coding Using Modified Three-step Search Algorithm for Block-matching Motion Estimation

Abstract
The major problem with fractal-based coding technique is that, it requires more computation at the encoding phase. Therefore, to reduce this computation, block-matching motion estimation algorithms are used. In this paper, we proposed a modified three-step search algorithm (MTSS) for block-matching motion estimation which consists of two cross-search patterns as a few initial step of search, and two cross-hexagon search patterns as a subsequent step of search that is used to search the center part of search window. Experimentations are carried out on standard video databases, i.e., football, flower, akiyo, coast guard, and traffic. The results of efficient three-step search and cross-hexagon search are compared with our proposed approach with respect to the parameters such as mean of absolute difference (MAD) and average search points per frame. Along with this our proposed approach, frame-based technique is used for fractal video coding evaluated on parameters such as compression ratio, PSNR, encoding time, and bit rate.
Shailesh D. Kamble, Nileshsingh V. Thakur, Latesh G. Malik, Preeti R. Bajaj

Designing and Development of an Autonomous Solar-Powered Robotic Vehicle

Abstract
In this paper, design and fabrication of a solar-powered robotic vehicle is presented. The energy for the vehicle is supplied by two solar panels of 5 W each. For efficient energy management, a charging system is designed. The charging is independent of the vehicle movement. Two batteries are used so that one is charging, while other is discharging. A charge controller is also designed so as to provide direct power supply to the connected load in case both batteries fail.
Pankaj P. Tekale, S. R. Nagaraja

SONIC: Stability of Nodes in Cooperative Caching for Wireless Network

Abstract
Most of the researches in wireless networks focus on data transfer through central node, and not much works have been done on data access, cache replacement, and stability of nodes. A general technique used to improve the performance of data access is cooperative caching. Cooperative caching has become one of the most exciting techniques for wireless environment. Cooperative caching allows sharing and coordinated cached data items which are stored in node’s local cache to achieve better performance. The main strength of this technique is lower node communication overhead and energy consumption as well as reduction in the query latency. Cooperative caching caches the frequently accessed data, the nodes do not always have to send request to server. But, high node mobility, constrained battery power, and limited wireless bandwidth may decrease the cache hit ratio and increase access latency or nodes in a region may fail to cache more hot data when a cooperative caching scheme cannot effectively distribute varied data to cache in that region. Nodes in wireless network may have similar task and share general interest, cooperative caching which allows the sharing and coordination of cached data among multiple nodes can be used to reduce the bandwidth and power consumption. In this paper, we propose the SONIC scheme; we find the stability of nodes, and the nodes which are more stable are being used first for searching the data, thereby increasing the network lifetime.
Vivek Kumar, Narayan Chaturvedi, Prashant Kumar

Advanced Energy Sensing Techniques Implemented Through Source Number Detection for Spectrum Sensing in Cognitive Radio

Abstract
The world of wireless technology is been one of the most progressive and challenging aspects for the users and providers. It deals with the wireless spectrum whose efficient use is of foremost concern. These are improved by the cognitive radio users for their noninterference communication with the licensed users. Spectrum holes detection and sensing is a dynamic time variant function which is been modified using the proposed source number detection and energy detection. Energy detection technique is implemented so as to compare the thresholds of the channels dynamically, and source detection method is used for predicting the number of channels where the energy detection is to be performed. The simulation results show the optimization and reduced probability of miss detection considering the change in threshold.
Sagarika Sahoo, Tapaswini Samant, Amrit Mukherjee, Amlan Datta

Dynamic Load Balancing for Video Processing System in Cloud

Abstract
Cloud computing is one of the more desired technologies in the recent times. It provides a wide range of services to users, common being the reliable virtual environment for storage and computation. With the demand for video content/video applications increasing rapidly over the years, real-time video streaming is becoming attractive with applications such as Video on Demand (VoD) and video conferencing. Streaming applications are resulting in increased traffic; thus, load on the network is increasing. Further worsening this situation is the user demanding for higher quality of video. Video application requires more storage and bandwidth resulting in a significant load on the network, and hence, a solution combining the cloud technology with multimedia is designed for balancing load in networks.
S. Sandhya, N. K. Cauvery

CT and MR Images Fusion Based on Stationary Wavelet Transform by Modulus Maxima

Abstract
In medical imaging, combining relevant information from the images of computed tomography (CT) and magnetic resonance imaging (MRI) is a challenging task. MR image carries soft tissue information that shows presence like tumor and CT image shows bone structures. For applications such as bioscopy planning and radio therapy, both kind of information is needed. This makes fusion problem more interesting and challenging. In this paper, we present an image fusion method based on stationary wavelet transform that decomposes source images into approximation, horizontal, vertical, and diagonal components. Coefficients of each of these components are combined using absolute maximum selection criteria separately. Inverse transformation results in a fused image. Also, the proposed method fuses images in presence of noise accurately. The performance of the proposed method is assessed visually and quantitatively. Entropy, fusion factor, and standard deviation are used as fusion performance measures.
Om Prakash, Ashish Khare

A Real-Time Intelligent Alarm System on Driver Fatique Based on Video Sequence

Abstract
The proposed work deals with development of an automatic system for drowsy driver detection using machine vision system. The system uses a small monochrome security camera that points directly toward the driver’s face and focuses the driver’s eyes. The video samples of drivers in drowsy and non-drowsy condition are obtained and stored in database. The video samples are converted to frames. Each frame is converted to binary images to track edges of eyes. An algorithm is developed to locate the eyes and its closure. After extracting the face area, the eyes are located by computing the average of pixels in horizontal area. Taking into account the knowledge that eye regions in the face present great intensity changes and the eyes are located by finding the significant intensity changes in the face. Once the eyes are located, measuring the distances between the intensity changes in the eye area determined whether the eyes are open or closed. The variation in intensity is plotted. Based on the distance between two valleys of the plot eyes, closure is detected. Once the closure is detected, fatigue is reported through a warning signal to the driver to it. The algorithm developed is unique to any currently published papers, which was a primary objective of our work. The performance of the work is reported for drowsy and non-drowsy driver’s samples in different environment.
Shamal S. Bulbule, Suvarna Nandyal

An Efficient Android-Based Body Area Network System for Monitoring and Evaluation of Patients in Medical Care

Abstract
Body area network (BAN) is a promising technology for monitoring different physiological parameters of the patients in real time. Particularly, when the BAN integrated with wireless technologies provides the complete medical infrastructure. Android is the most popular operating system in the smartphone. The wireless BAN combined with an Android-based smartphone offers a large functionality. Various medical parameters can be analyzed, stored, and visualized using the GUI of an Android application designed for the end user. The different sensors placed on the body of the patient acquire physiological data from patient. This acquired data is then, gone through signal processing and data analysis and results are sent to coordinator node. The physiological data are transferred, via bluetooth to an Android-based smartphone. The physiological parameters of the patients are continuously monitored by the system and if any variation occurred, it sends alert messages to the doctor. The alert is of two types, SMS alert and email alert. Using this alert system, the emergency situation can be handled effectively and the patient will get the medical care as soon as possible.
Lakshmana Phaneendra Maguluri, G. N. B. Markandeya, G. V. S. N. R. V. Prasad

An Efficient Kernelized Fuzzy Possibilistic C-Means for High-Dimensional Data Clustering

Abstract
Clustering high-dimensional data has been a major concern owing to the intrinsic sparsity of the data points. Several recent research results signify that in case of high-dimensional data, even the notion of proximity or clustering possibly will not be significant. Fuzzy c-means (FCM) and possibilistic c-means (PCM) have the capability to handle the high-dimensional data, whereas FCM is sensitive to noise and PCM requires appropriate initialization to converge to nearly global minimum. Hence, to overcome this issue, a fuzzy possibilistic c-means (FPCM) with symmetry-based distance measure has been proposed which can find out the number of clusters that exist in a dataset. Also, an efficient kernelized fuzzy possibilistic c-means (KFPCM) algorithm has been proposed for effective clustering results. The proposed KFPCM uses a distance measure which is based on the kernel-induced distance measure. FPCM combines the advantages of both FCM and PCM; moreover, the kernel-induced distance measure helps in obtaining better clustering results in case of high-dimensional data. The proposed KFPCM is evaluated using datasets such as Iris, Wine, Lymphography, Lung Cancer, and Diabetes in terms of clustering accuracy, number of iterations, and execution time. The results prove the effectiveness of the proposed KFPCM.
B. Shanmugapriya, M. Punithavalli

Backmatter

Weitere Informationen

Premium Partner

    Bildnachweise