A preferential digital–optical correlator optimized by particle swarm technique for multi-class face recognition

An optimized preferential digital–optical correlator is proposed for multi-class face recognition, where false rejection rate and false acceptance rate are improved by incorporating the information preferentially of both intra-class and other class face images for a given database during the synthesis of the correlator filter. Class compactness of both types of classes is made which makes the system more robust to distortion tolerance as well as misclassification. The optimization of trade-off parameters for both constrained and unconstrained type is carried out by particle swarm technique where particle vectors are considered as the correlator parameters to be optimized. Results on standard face databases have established that the performance of proposed correlator is better and more robust than the other existing classes of correlation filters. A digital–optical hybrid hardware architecture is developed and the performance for fast multi-class face recognition is compared with another existing hardware. The main contribution of the paper is related to the selection of optimum filter parameters and hardware testing in the two digital–optical hybrid architectures for multi-class face recognition.

► Modified correlation filter for multi-class face recognition is proposed.
► PSO based optimization of trade-off parameter is done.
► Hybrid digital–optical correlator architecture is suggested.
► FAR and FRR are minimized with maximization of IRR.
► Improved face recognition obtained under illumination conditions.