Intelligence surveillance system is based on computer vision. Throughautomatic analysis of the image sequences which was recorded in the camera, it canposition, detect and track the dynamic object. It can analyze and judge the object'sbehavior without human intervention. So intelligence surveillance system can notonly fulfill the traditional daily surveillance, but also have the ability to take place ofhuman's surveillance to deal with some accidents. With multi-camera intelligencesurveillance system can resolve some problems that exist in traditional surveillance:the shielded object detecting and tracking, irrelevant structure interference, lightsensitivity, suitable scene been limited and so on. So in the paper, the intelligencesurveillance system use multi-camera is more robustness and accuracy than singlecamera intelligence surveillance system.The paper exposes a new intelligence surveillance system which is combinedwith three-dimension reconstruct and artificial intelligence and builds a prototype ofthe intelligence surveillance system. In the paper, I also construct a new fastreconstruction method. With the method, first we need to find out the zone ofmoving object, then we can cut the zone off from photograph, at last we use the zoneof moving object to reconstruct the three-dimension object. With the help of themodel-based expert system, we can obtain the three-dimension object efficiently.Then we can put the three-dimension object into behavior analysis expert system, wecan analyze the behavior form the intelligence surveillance system. At last we testthe system and prove it can fulfill the motion detection and tracking requirementsefficiently. |