Video Active Security System Research And Construction

In the contemporary society, video surveillance system has been widely deployedin every field. It plays a critical role in preventing, frightening and striking crimes andbecomes a significant composition in the modern all-in-one security administrativemechanism. During the law enforcement process of public security, video surveillancesystem performs as a rather efficient assistant case-solving tool, and dramaticallyraises the success ratio of both solving cases and convictions. However, facing theproblem that how to make the video surveillance media contents contribute to judicialevidences, the importance of watermarking technology which works for the videoauthenticity emerges.This thesis is aiming at developing a new video watermarking system which hasthe ability to resist geometric attack, thereby making this new technology serves asone part of the video surveillance system functions, which means addingwatermarking singles when the video is recording. No matter what malignantgeometric attacks the video could suffere later on, the embedding water marks couldstill be examined and extracted correctly due to the fine integrity, in order to prove thecredibility of video tapes.This paper focuses on three-dimensional zoom stable feature point extractionwhich is based on Hessian Matrix, associated with minimum spanning tree, affinecovariant regions normalization and reverse normalization algorithm, as well as thelatest pseudo three dimensional DCT change and quantization index modulation, thenresearches and develops a new video water marking embedding and extracting systemwhich could utilized in distinguishing authenticity of video surveillance.The experiment results indicate that integrating algorithm with key technologybasically meet the design’s requirements. Although the attackers acquire the specificalgorithms, the videos still have no chance to be watermarking forged. Therebyguarantying the extremely high security of the video tapes.