Face Feature Extraction Algorithm Based on Wavelet Transform and CNN

To improve the accuracy and efficiency of face recognition algorithm, a convolution neural network(CNN) face feature recognition scheme based on wavelet transform is proposed in this paper. Firstly, the face image is decomposed into four regions with different frequencies and scales by wavelet transform. Then, the compressed image is transformed by discrete cosine transform, and the weighted distance is used for classification and recognition. The improved lightweight CNN is adopted in face recognition algorithm to effectively eliminate the noise. Based on MATLAB platform, the feasibility of such method is tested in the collected face image database. The simulation results show it has higher recognition rate and robustness, and better comprehensive performance compared with the traditional algorithm.