A novel Cosine–Sine hybrid chaotic map for secure UAV image encryption with dual Scrambling–Diffusion architecture

Unmanned Aerial Vehicles (UAV) are increasingly used for surveillance due to their mobility and capability to operate in hard-to-reach or hazardous environments, especially within industrial and critical infrastructure sites. UAVs are equipped with high-resolution cameras that enable them to capture detailed images which contain critical and sensitive information such as the structural design of buildings, the movement and operations of personnel or machinery, and the positioning of security systems. These sensitive images need to be protected so that only authorized people can access them, as they can be easily exposed to unauthorized access if not properly encrypted. In this paper, a novel one-dimensional Cosine Sine Hybrid (1D-CSH) Chaotic map and novel block-level dual scrambling-diffusion based UAV image encryption is proposed. The encryption process comprises four main stages. First, the input colour image is padded to ensure its dimensions are multiples of 8, facilitating division into 8 × 8 blocks. Next, the image is decomposed into its RGB channels and each channel is partitioned into 8 × 8 blocks. In the third stage, dual block scrambling is applied, where the scrambling phase is divided into two parts. In first part, pixels within each 8 × 8 block are scrambled via column transformations. In the second part, inter-block shuffling is performed, where blocks are rearranged according to indices generated by the proposed 1D-CSH chaotic map. Finally, the last stage is diffusion, where pixel values are modified using XOR operations combined with pseudorandom numbers generated by the Logistic Sine Cosine (LSC) chaotic map, enhancing pixel-level security and producing the final encrypted image. The proposed 1D-CSH chaotic map has been evaluated using bifurcation diagrams, Lyapunov exponent, Shannon entropy, and the NIST test for randomness, the result demonstrates that the map has strong chaotic behaviour with a Lyapunov exponent of 4.0. Experimental results confirm the effectiveness of the proposed scheme by achieving a Shannon entropy of 7.999 and a NPCR (Number of Pixels Change Rate) value of 99.61%, indicating high resistance to high resistance to statistical, differential, and brute-force attacks.