A Chosen-Ciphertext Attack for Chaos-Based Image Encryption

Image encryption is of utmost importance in modern times, especially with the wide range of applications in sensitive domains like defense, healthcare, space exploration, etc. Due to the bulky nature of image data, the encryption techniques are required to be lightweight, which, if not designed with due diligence, can compromise security. Cryptanalysis is a major tool to identify weaknesses in existing schemes and improve them. In this paper, we identify common weaknesses forming a class of chaos-based image encryption schemes vulnerable to chosen-ciphertext attack. Such weak schemes employ single or few rounds of encryption, each involving simple permutation/substitution operations resulting in poor diffusion. We exploit the oversimplified relationship between the plain image and cipher image pixels to design a General Chosen-Ciphertext Attack (GCCA) cryptanalyzing this class of schemes. The proposed GCCA uses three chosen-cipher images to completely recover any general [Formula: see text] targeted original plain image. Besides theoretically demonstrating full recovery of the plain image, we also use the deduction approach to show our attack’s efficacy by cryptanalyzing two existing chaos-based image encryption schemes (proposed by [ Ali et al. , 2023 ; Lone et al. , 2022 ], resp.) possessing identified weaknesses. Both of them over-rely on the chaotic properties of the utilized chaotic systems and other used mathematical primitives while overlooking the careful design of the respective schemes. Further, a general improved design for chaos-based image encryption schemes is proposed by mitigating the identified weaknesses. We also provide its security analysis and its demonstration by improvising the two cryptanalyzed schemes.