Study On Object Tracking Algorithm Based On Low Cost PTZ Camera

This paper is on the research of intelligent object tracking algorithm based on low cost PTZ camera. An efficient system which detects and tracks moving object robustly using low cost PTZ camera is proposed.The covariance of video sequences was calculated to detect moving objects. The number of frames to calculate the covariance changed automatically according to the scale of motion, so that the system worked well on all kinds of environments. At the same time, we stored the covariance in a spatial model, so that_motion detection of later frames could be enhanced by comparing current covariance with what was stored in the spatial model. The spatial model learned from the covariance of environment to detect source of interference. It not only eliminated many small motions that we were not interested in, but also improved the SNR performance. On the other hand, we stored the tracking information of objects in a special database, so that we can predict the motion on later occasion by checking out the history in the database. This made it possible to recognize the same object on another picture, and helped to keep the object in the center of camera. As a result of the database, the system worked well on situation with more than one moving object.We achieved the purpose of tracking object in real time, using special iterative algorithm optimized for PC and intelligent motion control algorithm for the PTZ camera. Experimental results showed that the algorithm worked robustly in various kinds of environments.