Abstract:

To improve the recognition rate and speed of face recognition, we propose a face recognition algorithm based on Gabor feature and projective dictionary pair learning. We first extract the Gabor features of the image at multiple scales and orientations, which shows significant robustness to the variations in expression, illumination and angle. The augmented Gabor features with reduced dimension are achieved through principle component analysis, which constructs a new training data to replace original training samples. Then a synthesis dictionary with reconstruction capability and an analysis dictionary with the capability of quickly obtaining representation coefficients are learned jointly during the training phase. The face is eventually identified by the reconstructed errors. Experimental results on the extended YaleB, ORL and AR database show that the proposed algorithm can get a high recognition rate and improve the recognition speed efficiently.

Key words: face recognition;gabor feature;synthesis dictionary;analysis dictionary