This paper develops an optimal privacy protection technique aimed at image security in video surveillance. Its technique is combined with various key problems such as de-identification, characteristics compressibility, recoverability, and preservation in one united architecture. Proposed technique developed a public stream and a secluded enduring error by distorting the secure sensitive portion. This study work takes into account both research areas and suggests a hybrid system for detecting and protecting human skin by using colour information under constantly changing lighting and environmental circumstances. With this motivation, proposed method is developed by considering two phases such as human skin detection and encryption which empowers the privacy of the surveillance video. The input image moved to the pre-processing stage of the Gaussian filter to remove noises present in the image. This process is obtained by using Enhanced Cipher Feedback Module Encryption (ECFME). In this process, the optimal key parameters are selected by the Modified Golf Optimization Algorithm (MGOA). Finally, the skin region is encrypted. The proposed methodology is assessed and evaluated by using performance measures to demonstrate the efficiency of the work compared to the conventional techniques.