A novel adaptive image privacy protection method based on Latin square

Abstract Recently, there has been a rise in image encryption methods using Latin squares. However, a significant portion of these methods relies on fixed key sequences, making them vulnerable to known plain image attacks (KPA) and chosen plain image attacks (CPA). To solve this, a novel adaptive image encryption scheme is proposed, which is based on the fractional order L$ \ddot{u} $ system and Latin squares. Specifically, the sum of each bit plane and the pixel value sum of the plain image are, respectively, used to select Latin squares and chaotic sequences in the proposed method, which is an effective approach for resisting the KPA and CPA. Moreover, using the fractional order L$ \ddot{u} $ system, the proposed method generates self-orthogonal Latin squares (SOLSs) to scramble the bit plane of the plain image in bit-level, followed by diffusing the scrambled image in pixel-level to boost the encryption’s robustness. The study compares the proposed method with state-of-the-art image encryption schemes and finds that it is superior in terms of usability, security, and efficiency. This method is highly promising for applications that require strong and efficient image encryption. 2020 MSC: 68P25