This paper analyzes the traditional Shannon-Nyquist sampling theorem, introduces the process of compressive perception theory and the key techniques of compressive perception in image sparse representation, design of measurement matrix and signal reconstruction, and explores the application of compressive perception in the field of image analysis and processing. Meanwhile, the image system imaging is constructed based on the compressive perception technique, and the process of wavelet packet subspace decomposition and reconstruction constructs the compressive perception image algorithm based on the optimal wavelet packet basis. The algorithm simulation results show that the minimum signal entropy is 16*4 in the minimum wavelet chunking way, at which the minimum values are -0.35, -0.04, -0.07, and -0.01, respectively.