Based On A Moving Background Of Moving Target Detection And Automatic Tracking

Considering the detection and tracking system generally need the artificially support, the paper researchers a new fully automated monitoring system. The paper mainly involves two parts:(1) detect the moving targets through some image processing algorithms; (2) control the camera to automatic track the moving target with the combination of the PC and the PTZ.In order to detect moving targets in the moving background, two aspects are needed to be done:firstly, match the background. The paper propose a robust background matching algorithm to complete the background's static conversion; secondly, introduce a detection algorithm - the three frames subtraction to detect the moving targets in the static background.There is a problem of tracking lag in the usually passive tracking algorithms, when the camera and the mobile move at the same time. In order to solve this problem, the paper combines the kalman filter algorithms and the Mean-Shift algorithms to complete the auto-tracking. Firstly, the kalman filter algorithms, which can initiatively the position of the object in the next frame is introduced to estimate the motion. According to the location of the next frame, the PC issue corresponding instructions to the PTZ. The PTZ controls the camera rotation. Next, determine a precise target location of the current frame using the Mean-Shift algorithms. Simultaneously, the Bhattacharyya similarity coefficient is introduced to determine whether the mobile has been obscured. This cycle continues.Based on the above ideas, a single-channel video surveillance system is set up under an experimental environment for programming and simulation after some algorithms are improved, integrated and optimized in this paper. Experiments show that the entire system achieves the desired results.