PREFACE

Digital images and videos are proliferating at an amazing speed in science, engineering and technology, media, and entertainment. With the huge accumulation of such data, the keyword search or manual annotation scheme may no longer meet the practical demand for retrieving the relevant contents in images and videos. Intelligent image search and video retrieval thus have broad applications in, for example, smart cities such as intelligent traffic monitoring systems, surveillance systems and security, and Internet of things such as face search in social media and on web portals. This book first reviews the major feature representation and extraction methods and the effective learning and recognition approaches that have broad applications in intelligent image search and video retrieval. It then presents novel methods, such as the improved soft-assignment coding, the inheritable color space (InCS) and the generalized InCS framework, the sparse kernel manifold learner method, the efficient support vector machine (eSVM), and the SIFT features in multiple color spaces. This book finally presents clothing analysis for subject identification and retrieval, and performance evaluation of video analytics for traffic monitoring. Specifically, this book includes the following chapters. Chapter 1 reviews the following representative feature representation and extraction methods: the spatial pyramid matching (SPM), the soft-assignment coding (SAC), the Fisher vector coding, the sparse coding and its variants, the local binary pattern (LBP), the feature LBP (FLBP), the local quaternary patterns (LQP), the feature LQP (FLQP), the scale-invariant feature transform (SIFT), and the SIFT variants. These methods have broad applications in intelligent image search and video retrieval. Chapter 2 first discusses some popular deep learning methods, such as the feedforward deep neural networks, the deep autoencoders, the convolutional neural networks, and the Deep Boltzmann Machine (DBM). It then reviews one of the popular machine learning methods, namely the support vector machine (SVM): the linear SVM, the soft-margin SVM, the nonlinear SVM, the simplified SVM, the efficient SVM (eSVM), as well as the applications of SVM to image search and video retrieval. Finally, it briefly addresses other popular kernel methods and new similarity measures. Chapter 3 first analyzes the soft-assignment coding or SAC method from