Researches On Key Technology Of Infrared Surveillance System

Infrared surveillance system plays very important role in the area of economy and construction of our country, and is widely used in the field of security control, intelligent traffic system, intelligent weapons, and so on. In order to resolve the inadequacies of existing system which cannot satisfy both the accuracy and real-time simultaneously, this paper mainly studies and discusses the key techniques of moving target detection and tracking in infrared surveillance system.First, image mosaics are studied for constructing a large field of view panoramic and acquainting with the scene of surveillance. Image registration is the base of image mosaics. Some of the known image registration and image blending algorithms are improved and also the simulation result is shown.The detection and tracking of moving targets is the key of infrared surveillance system. Two blueprints are put forward based on the methods of temporal difference and background subtraction on the research of moving object detection. Through the analyses of difference of adjacent images and their noise models, a method of automatic estimation of noise parameters is brought forward. For background subtraction, a new method of motion detection and background updating is developed based on the research of existing methods. Experiments show that the method is robust in dynamic variations. Then, an improved background subtraction algorithm based on Gaussian mixture model is proposed. On the research of moving object tracking, a new method of template updating is introduced after studying several relative algorithms. A Kalman filter is used to update the template dot by dot instead of renewing it as a whole, and make it adapt to changes in object orientation, illumination and occlusion.