Research On Recognition Algorithms And Alarm Based On Video Surveillance

With the rapid development of social and economic, people to work and life of the requirement of security is getting higher and higher. Video surveillance system (VSS) is an important part of guard and alarm system, has the strong comprehensive prevention capacity. VSS with its intuitive, accurate, timely, and information content rich and widely is used in various places of security and guard.Base on the characteristics of application for VSS in security field at home and abroad, the new methods of invasion criterion is proposed based on biometric identification for intrusion detection recognition alarm system, in this thesis. The methods of capture video are introduced base on USB2.0 by used two different development software. The characteristics of extracting moving object properties are analyzed by means of frame difference, background image difference and optical flow methods. Author puts forward a new method of intrusion features criterion according to the movement area ratio of width to high combining with sports trend and scope. The recognition algorithms are designed based on skin color model and human ambulation as criterion of gravity amplitude method. At last, a method of information fusion method is proposed based on rough set theory on human recognition alarm. The software is programmed with MATLAB and Visual C++ 6.0. Experimental result shows that methods can in real-time monitoring of real-time video in the meantime clips and save data frame, and can effectively distinguish people and other objects in motion, has high recognition accuracy and robustness. The VSS recognition algorithm which is proposed in this thesis has far-reaching significance in the field of security and guard.