Face Recognition Techniques and Analysis

This book presents standard as well as novel facerecognition methods. These methods utilize PrincipalComponent Analysis Linear Discriminant AnalysisIndependent Component Analysis Gabor WaveletsNeural Networks Hidden Markov Models GraphMatching etc. Emphasis is given to the popularEigenfaces algorithm which is presented analyticallyin detail and a framework is presented for itsexperimental evaluation. In addition this bookpresents a novel face recognition method that iscomputationally efficient and can be implemented as areal-time process. This method operates in quantizedblock histogram face spaces. Next a classificationalgorithm which inherently applies the optimumclassification measure in these spaces ismathematically derived. The development of thisalgorithm was motivated by the practical limitationsthat impair the performance of the Eigenfaces method.To overcome these limitations theoretical andexperimental statistical criteria are derived inorder to achieve high recognition rates. Thus anovel and potent face recognition framework ispresented along with other standard face recognitionmethodologies.