The Internet and its transmission routes have grown rapidly, making it simpler for attackers to obtain unwanted exposure to visual data. Even academics who gather photo collections for monitoring are impacted by privacy concerns. Existing techniques, though, are still vulnerable to assault. To lessen this possibility, a system that accepts data or photographs as input does preliminary processing and then uses machine learning techniques to produce a categorization output is being developed. This result is employed in identifying an acceptable encryption process that is utilized to safeguard the data, such as creating an encrypted share or employing various chaotic maps. By determining the encryption method employed, the decryption procedure aids in the reconstruction of the original data and enables an assessment of the system’s precision and degree of safety. This technology aims to make encrypted images more difficult for attackers to decrypt, hence increasing the safety of critical visual information. Multiple approaches are offered to be employed in the work that is being presented. Entropy and PSNR are used to examine the outcome.