The image is an important medium for information transfer. To ensure image security and improve encryption efficiency, a novel image encryption algorithm is proposed in this paper. Firstly, a new chaotic map is proposed by combining traditional chaotic maps. The SHA-256 algorithm processes the input image to produce chaotic parameters that generate random sequences associated with the plaintext. Secondly, the plain image is encoded using a fast-dynamic DNA encoding method to obtain the DNA matrix. Thirdly, a new DNA sequence operation method called DNA frame shift mutation is designed. The DNA matrix after performing the frame shift mutation transformation is decoded into a pixel matrix. Finally, coordinate matrices are constructed and the pixel positions are randomly selected for dynamic diffusion, which enables simultaneous scrambling and diffusion at the pixel level. The values of the information entropy are all greater than or equal to 7.9993. Experimental results show that the proposed algorithm is secure and it can be resistant to various attacks. Meanwhile, it has sound encryption efficiency.