An Image Encryption Method based on Pixel-level Diffusion and Bit Flip
Shiwei Zhang, Xin Lyu, Yunfei Zhang · 2021
With the rapid development of multimedia technology, digital images are increasingly used. These applications bring great convenience to users but increase the risk of information leakage. According to the granularity of operation, the existing image encryption methods can be divided into pixel-level and bit-level. The pixel-level method has high encryption efficiency, but does not break the linear relationship totally; the bit-level method has high security, but it needs to convert pixel value to 8 bits, which causes the low encryption efficiency. The current research on the hybrid encryption method performs pixel-level and bit-level operations in series. Although the methods can improve security to some extent, they still unsolve the inefficiencies problem. In this regard, we proposed a pixel-level bit-level hybrid diffusion image encryption method based on the traditional permutation-diffusion structure. The randomness is significantly improved by calculating the diffusion order through a double chaotic matrix. Meanwhile, we embedded bit flip into the diffusion process to further enhance the security of encryption. The experimental results show that our method achieves better results in five security metrics (pixel histogram analysis, key sensitivity, plaintext sensitivity, pixel correlation, and information entropy) compared with existing image encryption methods. Also, the time required for encryption is reduced by 6% compared to mainstream efficient encryption algorithms.