| Sumario: | The continuous advancement of communication technologies and data exchange over public networks has highlighted the crucially of protecting sensitive digital images and countering growing security threats. Traditional encryption methods such as RSA and AES face significant challenges when applied to digital images due to the large datasets and high correlation between adjacent pixels. This research presents a comprehensive overview of many digital image encryption algorithms that combine multidimensional chaotic systems (such as 2D Logistic-Sine-coupling maps) with dynamic DNA computing techniques. Another approaches combines hyperchaotic systems with compressive sensing to achieve simultaneous encryption and compression while providing visual security, also group of researchers proposed several ideas for encrypting digital images based on integrating Dynamic Spatial Chaos theory with modern models of Deep Learning and Edge Computing. Most of the proposed methodologies involve two main phases: permutation and diffusion, in the permutation phase, random sequences generated by the chaotic system are used to shuffle pixel positions across the entire image, breaking spatial correlation, while in the diffusion phase, pixels are converted into DNA sequences, and logical and arithmetic operations (such as addition, subtraction, and XOR) are applied based on DNA coding rules, significantly altering the pixel values. The results of experimental simulations and comprehensive security analysis showed that most of the proposed algorithms possess a very high level of security and a large key space, thus thwarting brute-force attacks. Statistical analysis also showed that the correlation coefficient between pixels in the encrypted image is close to zero, and the information entropy value is very close to the ideal value of 8. These algorithms demonstrated excellent resilience to differential attacks as measured by the NPCR and UACI values. Of course, every algorithm has its own advantages and characteristics depending on the application and research objective. Likewise, there are a number of limitations or weaknesses that these algorithms face, the most important of which are time and cost, as explained in the body of the research.
|
|---|