Deep learning based Image Compression, Encryption, and Recovery using Chaotic map and Compressive sensing
Rachuri Tarun, Priyanshu Aggarwal, Ashwini K M · 2025
Images are present in everything these days, from satellite photography to medical scans, and they have a profound impact on how people view and understand the world. The daily generation of an ever-increasing volume of image data has made it imperative to identify effective ways to convey these images. This research presents a novel picture recovery approach that makes use of Chaotic Maps for increased security and Compressive Sensing for effective compression. Compressive Sensing makes use of the sparsity seen in images to minimize the number of data points required. Chaotic Map is used in encryption, is also to guarantee the security of the compressed data. The Neural network model built to recover images is a hybrid of Conditional Generative Adversarial Networks (C-GANs), patch discriminator and U-net generator. This allows for high-quality image reconstruction. This approach is suitable for usage in healthcare and marine life.