Surveillance Video Encryption Technology Research Based On Foreground And Background Separation

With the development of computer technology, the popularization of broadband and the improvement of image processing technology, video monitoring is widely permeating the social from all walks of life. Especially, surveillance video in network transmission is vulnerable to suffer hostile attacks, which may causes citizen privacy in monitoring to be infringed. Therefore, surveillance video security encryption has a strong practical application value. Because of large amount of surveillance video data, it is a challenging task to ensure the real-time for encryption method. But it has far-reaching theoretical significance.This paper proposes a new surveillance video encryption method. This method takes into account the surveillance video is captured by a static camera and many similarities exist among video frames. First we select a few representative video frames to train a background recovery dictionary. Based on this dictionary, every video frame is separated to two parts, one is foreground region, another is background region. We only encrypt the foregrounds. Background regions are recorded by recovery dictionary and are not necessary to encrypt. Most of the foregrounds account for only a small proportion of the whole surveillance video. Therefore, needing to be encrypted parts are reduced and the encryption speed is improved. In order to guarantee the security of encryption and visual degradation effect, this paper has improved the image encryption algorithm which has a prominent encryption effect and transplanted it to the surveillance video encryption.In order to validate the method, we have do a lot of experiments based on both actual and simulated data. At the same time comparisons are done in all respects with the currently used encryption method. The experimental results have showed the effectiveness and feasibility of our proposed method.