Video key frame extraction correlated with feature selection and stage wise rejection based classifier for human face recognition

Abstract

Biometric recognition is a process, which determines an individual s activities and biological characteristics with the support of human face. Biometric systems are used to distinguish human face and access information about human behavior. Here, the recognition process comprises various stages, such as key-frame extraction, feature extraction, classification of relevant features and recognition of face images. Face recognition process is developed when there are variations in biological attributes and behaviors, both surrounded by different persons. Thus, biometric systems are mostly used to overcome the challenging issues, in recognizing human face image. In recent times, several works have been considered to enhance human face recognition accuracy with reduced time. In the existing face detection models, key-frames with relevant features are not extracted to provide higher recognition accuracy. Here, haar wavelet is used to detect the face regions and comparisons are made based on the threshold value. However, action video region frames have not been considered. The existing multimodal processes are not efficient for achieving higher recognition rate. On account of the presence of high dimension features, the biometric recognition system is extremely complex. In order to overcome the above mentioned issues, different methods namely KEWI scheme, TA-MMFS method, VQOS-RC model and OGOA-FS model have been developed with various optimal feature selection processes and classifier for biometric recognition. newline