A face recognition algorithm based on compressive sensing and wavelets transform

In face recognition, image brings great inconvenience to the hardware because its high degree of redundancy and a largequantity of data. This paper introduces a theory of Compressive Sensing(CS) for Face Recognition (CSFR) that a DWT is applied to the dictionary which created by all training samples, then the image is processed by CS in wavelet domain. The reconstruction is computed withOrthogonal Matching Pursuit(OMP) algorithm, and the residual(the distance between the reconstruction vector and the training vector) determines the class of thetest data. The computer experiment on ORL database shows that the CSFR algorithm based on DWT (DWT-CFSR) performs more robust and effective in face recognition than SRC and PCA algorithms.