Scrambling for Privacy Protection in Video Surveillance Systems
IEEE Transactions on Circuits and Systems for Video Technology
| Hi-index | 0.00 |
This paper defines a new methodology for trustworthy video surveillance databases, which fulfills four crucial properties: high accuracy, reversibility, real-time performance and information security. We detect the regions of interest (e.g. faces of people), which are protected by means of the alteration of the coefficients of the compressed video stream. The proposed protection guarantees the property of reversibility, since the original coefficients can be restored. Specifically, we elaborate on the definition of the Protection Stream, which is required for both protecting and unprotecting a video sequence. We propose a procedure for securely generating a protection stream for each group of pictures, thus avoiding the large unsecure streams required if the protection stream was generated for each video file. The proposed model has been implemented and tested, and the results confirm the real time performance of the system developed while keeping the aforementioned crucial properties.