Secure ML Based Image Encryption Framework with Recursive Splitting and Linear Feedback Shift Register

With the development of advanced communication ensuring the secure transmission of images has become more and more important. This research suggests a novel approach for digital image encryption and decryption that combines Linear Feedback Shift Register (LFSR) encryption, recursive image splitting and shuffling in order to increase security. Test image is divided into blocks and each block is encrypted using a different LFSR key. The encrypted blocks are then joined to create a single encrypted image that is ready to be sent. The receiver uses a machine learning model that has been pre-trained to identify the specific LFSR key used for each block after breaking the image back up into its component blocks. This allows for precise decryption. The viability of the encryption process is assessed using entropy analysis, histogram comparison and pixel relationship etc. The results show that the encrypted images have a notable disruption of spatial relationships, a high degree of unpredictability, and significant interference with spatial relations. Furthermore, evaluation is done on the performance of multiple Support Vector Machine (SVM) models with different kernel functions. The results indicate that the SVM with a radial basis function (RBF) kernel performs best in terms of key identification accuracy. This thorough analysis demonstrates the suggested encryption method's resilience and its potential for safe image transfer across a range of applications.