A Compressed Sensing and Image Encryption based on Adaptive Rossler Hyper Chaotic Encryption for Tamper Localization and Recovery Model

Objectives: A novel image encryption with a tamper localization model is suggested in this paper. Focusing on tamper localization has the efficiency to accurately protect the integrity and authenticity of images. In order to improve the quality of the image, the tampering localization plays a crucial role in locating the unauthorized changes in the images. Methods: Initially, the raw images are collected from the standard dataset. Further, the collected images are given to the Adaptive Dual-Tree Complex Wavelet Transforms (ADTCWT), which divides the input image into low and high-frequency images. Here, the attributes in the ADTCWT are optimized by the Improved Social Engineering Optimizer (ISEO). The obtained high-frequency images are encrypted using the Arnold Scrambling. Here, the sparse representation is employed to do the Compressive Sensing (CS) in the high-frequency images. Similarly, the low-frequency images are encrypted by the Arnold Scrambling, and also, the watermarking process is carried out on the low-frequency image. The processed low and high-frequency images are fed into the Adaptive Rossler Hyper Chaotic Encryption (ARHCE) model for the encryption process. Moreover, the same ISEO is employed for tuning the attributes in the structure. After that, the inverse ADTCWT is used to perform the inverse decomposition to get the original images in encrypted form. Also, the encrypted images are provided to ARHCE for image decryption. Then, the decrypted images are given to the ADTCWT for decomposition to generate low and high-frequency images. The watermark extraction is carried out on the low-frequency image. Hence, the tamper localization and the recovery process are carried out on the low-frequency image. Afterward, the high-frequency image is subjected to the inverse CS, and it is decrypted by the Arnold Scrambling. The decrypted low-frequency and high-frequency images are fed into the inverse ADTCWT to get the original image. Results: Finally, the performance of the developed model is evaluated by comparing it with the existing techniques in terms of various measures. Comparing PSNR measures in the developed framework, the performance shows 29.2%, 20.4%, 10.4% and 35.8% over DWT, Chaotic map, NSCT, and RHCE at varying image sizes of [Formula: see text], respectively. Conclusion: By validating these performance measures, the developed model is widely applicable in large scale analysis and also it provides improved tamper localization performance. This extensive experiments shows the developed model provides robust performance than the state-of-the-art-methods.