Abstract:  To utilize the Gabor feature for efficient face recognition‚a new Gabor feature extraction method was pro-posed．Firstly‚a criterion with discriminating power was used to evaluate the classification ability of the Gabor coeffi -cients‚and the Gabor coefficients most favorable to the recognition were selected to construct new Gabor features．Then‚the classification abilities of different local positions in the face were evaluated‚and feature points were chosenat the positions with the strongest discriminating power．Finally‚the adaptability of the new Gabor feature to illumina-tion and expression was tested．Experiment results show that‚with small samples‚the new feature extraction method can effectively increase the recognition rate‚reduce the number of features‚and is of certain adaptability to the envi -ronments．

Key words: face recognition, Gabor feature, feature extraction