5D chaotic map-based image encryption trade-off analysis on various stages of encryption

The protection of multimedia data in unsecured networks has become increasingly critical due to the widespread use of images across various domains. Traditional encryption methods primarily focus on maintaining the confidentiality of image data. This paper presents a novel image encryption technique based on a five-dimensional (5D) memristive hyperchaotic system, which generates highly unpredictable random sequences for the diffusion process. Additionally, a linear feedback shift register (LFSR) is employed as a hardware-efficient confusion mechanism, ensuring seamless implementation on resource-constrained devices. Extensive experimental evaluations were conducted on various image datasets, incorporating statistical analyses such as entropy (7.99732), peak signal-to-noise ratio (PSNR) (8.77468 dB), histogram variance, and Chi-square test analysis. Differential analyses, such as the number of pixels change rate (NPCR) (99.63152%) and unified average changing intensity (UACI) (33.50686%), show that the proposed encryption scheme works even better. The robustness of the approach is demonstrated by its resistance to noise, cropping, and chosen-plaintext attacks. The use of a 5D chaotic system significantly expands the key space, and the unpredictability of the generated keys was verified using the NIST 800–22 statistical test suite. To evaluate hardware feasibility, the encryption algorithm was implemented on an Intel Cyclone IV FPGA, utilizing 2012 logic elements and consuming 135.4 mW of power, achieving a throughput of 655.36 Mbps. The results confirm that the proposed method is well-suited for secure, real-time image encryption in embedded systems and resource-constrained environments.