The security of image data is of great significance for protecting personal privacy, commercial secrets, and national security. As a crucial means of safeguarding image security, image encryption has garnered extensive research interest. This paper proposes an image compression and encryption algorithm based on Multi-Objective Particle Swarm Optimization (MOPSO), chaotic mapping, compressed sensing (CS), and DNA encoding. First, the pseudo-random sequences generated by the Two-Dimensional Cosine–Sine Interleaved Chaotic Map (2D-CSIM) are used to construct the measurement matrix for compressed sensing. Then, the image pixels are diffused using a novel four-dimensional conservative chaotic system and DNA rules. The purpose of MOPSO is to optimize the entropy of the ciphertext image and the correlation between adjacent pixels. Finally, the optimal key is returned. Experimental results and extensive security analyses demonstrate that the proposed scheme exhibits excellent performance, robustness, and strong resistance to various attacks.