Optimization of Image Zernike Moments Shape Feature Based on Evolutionary Computation

The image shape feature can be described by the image Zernike moments. In this chapter, the authors point out the problem that the high dimension image Zernike moments shape feature vector can describe more detail of the original image but has too many elements making trouble for the next image analysis phases. Then the low dimension image Zernike moments shape feature vector should be improved and optimized to describe more detail of the original image. Therefore, the optimization algorithm based on evolutionary computation is designed and implemented in this chapter to solve this problem. The experimental results demonstrate the feasibility of the optimization algorithm.

Thresholding Selection Based on Fuzzy Entropy and Bee Colony Algorithm for Image Segmentation

Image thresholding segmentation based on Bee Colony Algorithm (BCA) and fuzzy entropy is presented in this chapter. The fuzzy entropy function is simplified with single parameter. The BCA is applied to search the minimum value of the fuzzy entropy function. According to the minimum function value, the optimal image threshold is obtained. Experimental results are provided to demonstrate the superior performance of the proposed approach.

Review of Sparsity Known and Blind Sparsity Greedy Algorithms for Compressed Sensing

Sparsity level is crucial to Compressive Sensing (CS) reconstruction, but in practice it is often unknown. Recently, several blind sparsity greedy algorithms have emerged to recover signals by exploiting the underlying signal characteristics. Sparsity Adaptive Matching Pursuit (SAMP) estimates the sparsity level and the true support set stage by stage, while Backtracking-Based Adaptive OMP (BAOMP) selects atoms by thresholds related to the maximal residual projection. This chapter reviews typical sparsity known greedy algorithms including OMP, StOMP, and CoSaMP, as well as those emerging blind sparsity greedy algorithms. Furthermore, the algorithms are analysed in structured diagrammatic representation and compared by exact reconstruction probabilities for Gaussian and binary signals distributed sparsely.

A Survey of Human Activity Interpretation in Image and Video Sequence

In the past, a large amount of intensive research has been dedicated to the interpretation of human activity in image and video sequence. This popularity is largely due to the emergence of the wide applications of video cameras in surveillance. In image and video sequence analysis, human activity detection and recognition is critically important. By detecting and understanding the human activity, we can fulfill many surveillance related applications including city centre monitoring, consumer behavior analysis, etc. Generally speaking, human activity interpretation in image and video sequence depends on the following stages: human motion detection and human motion interpretation. In this chapter, the authors provide a comprehensive review of the recent advance of all these stages. Various methods for each issue are discussed to examine the state of the art. Finally, some research challenges, possible applications, and future directions are discussed.

Adaptive Edge-Preserving Smoothing and Detail Enhancement for H.263 and H.264 Video

For removing undesirable artifacts in video coding, a large number of filtering methods have been proposed as post-processing and in-loop processing. This chapter proposes a pre-processing method of motion-adaptive edge-preserving smoothing and detail enhancement for H.263 and H.264 video, in which temporal and spatial edges are used to define Region Of Interest (ROI). In the proposed pre-processing method, trilateral filtering with three types of weights (domain, range, and temporal weights) is used for smoothing non-ROI region while preserving temporal/spatial edges. In the proposed pre-processing method, the temporal weight preserves temporal edges within ROIs and smoothes details within non-ROI. The proposed pre-processing method can make video coding more efficient under the restricted bit-rate condition. The parameter values for weight functions of a trilateral filter are selected depending on the classification of motion and edginess, and proper filtering is performed with adaptive parameter values. Experimental results with a number of H.263 and H.264 test sequences show the effectiveness of the proposed method in terms of the visual quality, the peak signal-to-noise ratio, and the mean opinion score.

Image Denoising via 2-D FIR Filtering Approach

Image denoising has received much concern for decades. One of the simplest methods for image denoising is the 2-D FIR lowpass filtering approach. Firstly, the authors make a comparative study of the conventional lowpass filtering approach, including the classical mean filter and three 2-D FIR LowPass Filters (LPF) designed by McClellan transform. Then an improved method based on learning method is presented, where pixels are filtered by five edge-oriented filters, respectively, facilitated to their edge details. Differential Evolution Particle Swarm Optimization (DEPSO) algorithm is exploited to refine those filters. Computer simulation demonstrates that the proposed method can be superior to the conventional filtering method, as well as the modern Bilateral Filtering (BF) and the Stochastic Denoising (SD) method.

Intelligent Video Authentication

With the innovations and development in sophisticated video editing technology, it is becoming increasingly significant to assure the trustworthiness of video information. Today digital videos are also increasingly transmitted over non-secure channels, such as the Internet. Therefore, in surveillance, medical, and various other fields, video content must be protected against attempts to manipulate them. Video authentication has gained much attention in recent years. However, many existing authentication techniques have their own advantages and obvious drawbacks; we propose a novel authentication technique that uses an intelligent approach for video authentication. This book chapter presents an intelligent video authentication algorithm using support vector machine, which is a non-linear classifier, and its applications. It covers both kinds of tampering attacks, spatial and temporal. It uses a database of more than 4000 tampered and non-tampered video frames and gives excellent results with 95% classification accuracy. The authors discuss a vast diversity of tampering attacks, which can be possible for video sequences. Their algorithm gives very good results for almost all kinds of tampering attacks.

Dimension Reduction of Local Manifold Learning Algorithm for Hyperspectral Image Classification

This chapter addresses the problems in hyperspectral image classification by the methods of local manifold learning methods. A manifold is a nonlinear low dimensional subspace that is supported by data samples. Manifolds can be exploited in developing robust feature extraction and classification methods. The manifold coordinates derived from local manifold learning (LLE, LE) methods for multiple data sets. With a proper selection of parameters and a sufficient number of features, the manifold learning methods using the k-nearest neighborhood classification results produced an efficient and accurate data representation that yields higher classification accuracies than linear dimension reduction (PCA) methods for hyperspectral image.

Audio and Visual Speech Recognition Recent Trends

This chapter focuses on a brief introduction on the origins of the audio-visual speech recognition process and relevant techniques often used by researchers in the field. Brief background theory regarding commonly used methods for feature extraction and classification for both audio and visual processing are discussed with highlights pertaining to Mel-Frequency Cepstral Coefficient, and contour/geometric based lips feature extraction with corresponding tracking methods (Yingjie, Haiyan, Yingjie, & Jinyang, 2011; Liu & Cheung, 2011). Proposed solution concepts will include time derivatives of mel-frequency cepstral coefficients for audio feature extraction, Chroma-colour-based (YCbCr) Face segmentation, Feature Point extraction, Localized Active Contour tracking algorithm, and Hidden Markov Models with Vitebri algorithm incorporated. Information contained in this chapter focuses on being informative for novice speech processing candidates but insufficient mastery knowledge. Additional suggested reading materials should assist in expediting field mastery.