Research On The Application Of Compressive Sensing In Video Forgery Detection

With the progress of digital multimedia technology, the multimedia equipments, such as digital cameras and digital video recorders, have become popular. Meanwhile, a lot of video editing software with powerful function has been widely used, so nonprofessional people are able to tamper with a video without difficulties. So it is becoming increasingly important and urgent to authenticate the veracity and the integrity of videos, which has already become the research hotspot in recent years.Because of the complexity of the video, such as large data volumes, high-dimensional and nonlinearity, it is becoming particularly important to reduce the data quantity, the detection complexity and the computation time."Compressive Sensing" is a new theory for signal processing proposed by Candes and Donoho et al. Applying this theory to video forgery detection effectively can reduce the computing complexity, improve the detection accuracy and robustness. The main work of this paper has following aspects:1. According to Based on the theory of Compressive Sensing, we present an improved algorithm for image reconstruction based on Smoothed-l0and Dual-Tree Wavelet Transform to save the execution time and improve the quality of reconstruction. In this algorithm, the Dual-Tree Wavelet Transform is used for image sparse and Random Hadamard matrix is used to measure the data obtained from Dual-Tree Wavelet Transform. During image restoration, the Smoothed-l0algorithm is selected for reconstructing the ima(?) he experiment results show that the algorithm improves the execution speed and the quality and robustness of the reconstructed image.2. Aiming at the forgery of moving foreground removed from static background, we propose a video forgery detection algorithm based on compressive sensing. In this algorithm, the features of the difference between frames are obtained through K-SVD (k-Singular Value Decomposition), and random projection is used to project the features into the lower-dimensional subspace, which is clustered by k-means. After obtaining the tampered traces of single frame, the algorithm combines and outputs the detection results with’or’operator. The experimental results show that our algorithm has higher detection accuracy and better robustness than that of the previous algorithms.3. Aiming at the forgery of copying and pasting some existing contents to another disjoint region in the same frame, we put forward a video tamper detection algorithm based on SIFT and compressive tracking. In this algorithm, SIFT is used to extract the features of video according to inspection interval parameter, and the first tampered frame and the tampered area are selected. And then the compressive tracking algorithm is used to study this tampered area and locate the suspicious area of subsequent frames. Finally, the similarity of suspicious areas is calculated and the tampered areas are confirmed. The experimental results prove that the algorithm proposed has better effectiveness, robustness and higher efficiency than previous algorithms.The innovative points of this paper mainly include:1. The algorithm combines the Dual-Tree Wavelet Transform with Smoothed-l0is proposed for the image reconstruction using compressive sensing, which can improve the quality and robustness of reconstructed image as well as the execution speed.2. Applying the compressive sensing theory to video forgery detection in which the moving foreground was removed from background, not only ensures the accuracy and robustness of the algorithm, but also reduces the amount of processing data greatly.3. For the detection of the video forgery by copying and pasting some existing contests to another disjoint region in the same frame, we propose a new algorithm combining the SIFT with compressive tracking, which greatly improves the accuracy and execution speed on the premise of algorithm robustness.