Digital Video Source Identification And Forgery Detection Technology For Safe City

Large-scale video surveillance systems support safe city,snow project,etc.and become an effective means of urban management and social governance.Aiming at equipment counterfeiting and video forgery for massive distributed cameras lead to severe information security challenges for large-scale video surveillance systems.This thesis focuses on video traceability and forgery detection for large-scale video surveillance systems.The specific work is as follows:Monitoring video source recognition results in low rate and low efficiency due to improper extraction of video features.Algorithms based on PRNU statistical features are difficult to achieve an ideal recognition rate.In this thesis,high-order wavelet statistics(HOWS)is applied to the representation of PRNU noise,and a representation method of video features is proposed.On this basis,a video device fingerprint extraction algorithm and a video traceability algorithm combining HOWS features and video frame statistics are proposed.The experimental results show that the proposed algorithm has a high recognition rate of 96.375%and anti-counterfeiting ability.It effectively solves the problem of low accuracy of monitoring video source identification,and it can meet the needs of the monitoring system to effectively prevent spoofing attacks.Aiming at the problem of illegal tampering of surveillance video,a passive forensic method based on key-point for video forgery detection is proposed.This method mainly solves the frame copy-mobile tampering problem of video,which includes three parts:video frame key-point extraction algorithm,suspicious frame research algorithm and repeated frame detection algorithm.For the extraction of key-points of video frames,the efficiency of video tampering detection algorithm can be greatly improved by using efficient ORB algorithm.Frame matching is performed using a Random Sampling Consistency Algorithm(RANSAC)to detect repeated frames in the video.It achieves the goal of efficiently and quickly detecting video tampering and solves the problem of monitoring video forgery attacks.The experimental results show that under the premise of ensuring 100%accurate detection and location of the tampered frame,the efficiency of the algorithm is 10 times higher than of the video forgery detection algorithm based on SIFT key-points,and the average detection time is 212.75ms.Based on the above proposed technologies and method,the safe city surveillance video security management platform is designed and implemented.Combined the characteristics of video feature extraction algorithm with the characteristics of the video surveillance system,and the surveillance video traceability algorithm and video forgery detection method proposed in this thesis are applied to the platform.The platform is deployed in the public security system to verify the feasibility of the proposed algorithm.