<p>As the key carrier of information transmission and storage, the security of digital images is facing severe challenges. In view of the unique advantages of chaotic system, such as randomness, unpredictability and high sensitivity to initial conditions, the image encryption algorithm based on chaos theory became a research topic, which provided an effective solution to improve the security of image data. In this paper, a hidden attractor chaotic system based on memristor model is successfully constructed, and its unique dynamic characteristics are revealed through comprehensive qualitative and quantitative analysis: the system has no equilibrium point and its largest Lyapunov exponent reaches 1.59, which highlights its complex and rich chaotic behavior. In addition, a chaotic image encryption scheme is proposed by using this new system, which skillfully combines pixel-level and bit-level scrambling and diffusion technologies. Security analysis shows that the information entropy of the scheme is 7.9998, the number of pixels change rate (NPCR) and the unified average changing intensity (UACI) are 99.61% and 33.53% respectively, and the key space is 10⁹⁰, which can effectively resist common attacks such as statistical analysis, differential cryptanalysis and brute force cracking, and fully proves the effectiveness and reliability of the scheme.</p>