In the face of massive face image recognition, traditional feature extraction methods are difficult to extract effective features, resulting in low face recognition accuracy.A robust face feature extraction algorithm is proposed, which uses the deep convolution sparse self-encoding network to automatically learn the features of the face that are rich and highly recognizable.This method integrates the convolution operation into the self-encoding network, and adds the sparse idea to form a deep convolution sparse autoencoder (HDCSAE) ;the network automatically extracts the high-level robust features of the massive face image, and uses the extracted features as the input of the SVM classifier to obtain the classification result.This method is tested under the FERET face database, and the recognition rate reaches 99.47%, which is better than that of the traditional face recognition method based on extracting artificially defined features.