Aiming at addressing the problems of dimensionality disasters and feature redundancy in Gabor wavelet extraction, a face feature extraction method based on wavelet transform and multifeature fusion coding is proposed. The proposed method uses a 2D-Gabor wavelet to extract normalized input image feature information to obtain Gabor features at different scales and directions. For each feature image, a multifeature fusion coding model based on the Gabor wavelet is used to extract the image L-F (Local Gradient Coding-Fusion) features. Finally, a histogram is used to count the image features to select the appropriate number of blocks, and the information entropy is used to multiply all sub-block images by their respective weighting coefficients to obtain the final face features, and use training samples for the Euclidean distance to set the confidence interval and identify it. The experimental results show that the proposed method has better performance than other feature extraction methods and can demonstrate good robustness under the influence of different poses and complex illuminations.