A new multiple-chaos image encryption algorithm based on block compressive sensing, swin transformer, and wild horse optimization
Qutaiba Kadhim, Waleed A. Mahmoud Al‐Jawher
Abstract
Due to remote working and learning, the quantity of images communicated on public networks has significantly increased in recent years. The need for image encryption algorithms with appropriate capacity and high efficiency is growing due to the limitations of traditional image encryption algorithms. This paper proposes a multiple-chaos algorithm based on block compressive sensing (BCS), swin transformer (ST), and Wild Horse Optimization (WHO) for encryption. First, the Discreet Wavelet Transform (DWT) will be applied to the image. Then, the image confusion will be performed using the FAN Transform. Second, The image will be divided into 16 blocks each block is 64×64 pixels, and each block is compressed using the Hadamard Matrix. All blocks are masked together and quantified in the contents of the image. Next, a diffusion process will be applied to image pixels through the chaos numbers, which were generated using the Swin Transformer (ST). As well as The Wild Horse optimization (WHO) was used as a scrambling tool to obtain the encrypted image. Finally, block compressive sensing (BCS) will be applied to increase the processing speed and security. From testing the proposed algorithm it was shown that it is efficient and robust enough to fend off numerous assaults, as shown by actual trials. The average information entropy in simulation trials was 7.9749, the average for NPCR was 99.5453%, the PSNR average was 34.0248 and the average for UACI was 33.4224%. The data indicates that the method is robust enough to withstand multiple attacks. Higher values of the average measurement confirm the security of the encryption algorithm.