An image compression-encryption algorithm based on BP neural network optimized with fireworks algorithm

With the rapid growth of the volume of image data, the burden of storage and transmission has increased dramatically. However, current algorithms face challenges in balancing compression efficiency with security. In this paper, a novel algorithm for image compression and encryption utilizing an optimized BP neural network is proposed. First, the fireworks algorithm (FWA) is employed to optimize the initial weights of a BP network, bolstering image compression efficiency and reconstruction quality. Second, a variable-parameter chaotic system is constructed by combining Logistic-Tent and Chebyshev maps. Meanwhile, the pseudo-random sequences are generated by a variable-parameter chaotic system for scrambling and diffusion. Subsequently, a multi-level global-local scrambling mechanism is introduced to disrupt the statistical properties of the image. Finally, a Gray Code-based mutation-diffusion strategy is implemented to achieve encryption. Experimental results indicate that the encryption algorithm exhibits strong robustness in terms of histogram analysis, information entropy, key space, and attack resistance. At identical compression ratios, the proposed scheme proves superior to BP neural networks, delivering higher reconstruction quality for high-fidelity image recovery.