To solve the problem that present recognition accuracy of optical image is not high, the method of image feature extraction based on wavelet space feature spectrum entropy is studied deeply. According to the principle of energy is equal before and after the wavelet transform, the matrix of wavelet energy mode is constructed, and the singular value decomposition is done to the matrix, then the feature spectrum entropy of singular value is as the image feature. At the same time, the back propagation neural network is also used for image recognition. The experimental results show that the proposed image feature extraction method can obtain higher accuracy of image recognition, and the effectiveness of the method is also proved.