On the divide-and-conquer attack of a plaintext related image chaotic encryption scheme

The typical categories of chaotic ciphers are mainly divided into plaintext-independent type and plaintext-related type. Due to the existence of equivalent key in the plaintext-independent type, it can usually be deciphered by simply using chosen-plaintext attacks. Recently, some plaintext-related chaotic ciphers have been proposed for improving the security performance. However, it is found that under the premise of predictability and reconfiguration of plaintext-related quantities, chosen-plaintext/ciphertext attacks can still remain valid. But if the above-mentioned conditions do not hold true, it is not possible for chosen-plaintext/ciphertext attacks to decipher the generalized plaintext-related chaotic ciphers. Therefore, it is necessary for cryptanalysts to discover some new analytical approaches. To further explore the problem, this paper proposes initially the divide-and-conquer attack on this type of plaintext-related chaotic ciphers,and takes the cryptanalysis on the latest similar scheme as an example to illustrate the main idea. Armed with the divide-and-conquer strategy, the related pixels are deciphered firstly. Then, the remainder pixels can also be deciphered followed by the chosen-plaintext attack, difference method, S-box strategy, and removing diffusion effect. Meanwhile, an improved scheme is also given to eliminate potential flaw, which have no need of additional plaintext information for decryption, but also can resist against divide-and-conquer attack. Theoretical analysis and numerical experiments have confirmed the feasibility of the method.