Abstract
Privacy is one of the great bases of our civilization and is important for many societal purposes. Therefore, in this paper, a data hiding technique was proposed to improve privacy in the video surveillance system and it can recover all the original scenes using a securely key for the authorized user. The original video frames that hold clear faces are embedded in a blurred version of frames using wavelet coefficients and an embedding function. That is, each blurred frame holds a secret frame that represents the non-blurred version. The proposed method was implemented using MATLAB, the computer vision toolbox and the frontal multiple faces dataset. Results captured by the PSNR metric show an acceptable visual quality for retrieved secret frames.