Remote sensing images are increasingly used in defense, agriculture, environmental monitoring, urban planning, and other fields, playing an irreplaceable role. Remote sensing images often contain sensitive geographic and strategic information. Ensuring their security during storage and transmission is significant for protecting data integrity and confidentiality, preventing information leakage, and maintaining national security. We propose a DNA image encryption method based on metaheuristic algorithm optimization. First, we improve the sine–cosine algorithm (SCA) and propose the LSCA-DE algorithm, which significantly improves the accuracy and convergence ability of the algorithm. Benchmark tests and statistical tests prove the effectiveness of the improvement; secondly, we use LSCA-DE to optimize the DNA encryption algorithm and propose the LSCA-DE-DNA image encryption algorithm; finally, LSCA-DE-DNA is used for Landsat-5 and Sentinel 2 remote sensing satellites. The experimental results show that the algorithm has well anti-clipping and anti-noise capabilities.