A survey on image cryptography mechanisms: hitherto, and future directions

With the exponential growth in the multimedia data such as images, videos, or audios, the security and confidentiality of the information has emerged as one of the challenging problems. In particular, image data security during communication raises several concerns, where attackers can intercept the communication channel and data security can be breached. Thus, image cryptographic algorithms are a viable solution to secure the images while communication between sender and receiver. Though, there has been significant progress in the field of image cryptography, security and performance of these algorithms is still a challenge to the research community. In this article, we expound a detailed taxonomy-based analysis of the existing state-of-the-art (SOTA) image cryptography techniques. Initially, we devised a novel taxonomy for broad categorization of the image ciphers. Later, the entire survey is organized by considering our own taxonomy for a clear and systematic structure. The analysis of various image ciphers is on the basis of three important aspects such as confidentiality, integrity and authenticity (CIA). Besides, we also illustrate various performance evaluation protocols and cryptanalysis methods that are frequently employed for analyzing the efficacy of the image cryptography algorithms. In addition, we explored the modern methods such as deep convolutional neural networks (DCCN) that are also frequently employed for image cryptography. This study exposes several open research challenges in this active field of image security and also provides future perspectives to the new investigators. One of the imperative challenges is to design lightweight image ciphers with ideal security characteristics. Moreover, authenticated image ciphers is an additional critical parameter for image ciphers, which requires attention of research community.