Study Of The Video Monitoring And Automatic Tracking System Based On ARM9

In the usual video monitoring system, the PC machines or the Industry Controlling Computer is used as the controlling core, because of the complicated calculate in the image processing and the needs of the high speed CPU. But they integrate too many functions which aren't relative to the monitoring system, and they costs much. In the system which is designed in this paper, the S3C2410 ARM microprocessor was used as the controlling core, on which the reduced Linux OS runs, to achieve the purpose of reducing the resources. The common monitoring systems need the human's assistance, aim at this problem, so the paper studied the arithmetic of auto-tracking of mobile in the video frames and the automatic controlling method of stepping motor of the pan-tilt, and realized entire automation of the monitoring system without the human.When the video and the mobile move at the same time, it is hard to realize the image recognization and the controlling of the step motor. The paper realized it as following: When the mobile is detected, the video will be stopped turning, and collect some continuous images to analyze them, and the relevant instruction will be sent to the pan-tilt to make it track the mobile to turn some angles and stop. And repeat it, and the images won't collected when the pan-tilt is turning.In the mobile's recognizing, the improved frame subtraction method based on the ratio between the pixels is used, and the recongnition errors is reduced especially in the condition of the low value of the pixels' grey in the images, improving the usual frame subtraction method; In the choice of the number of the frames of the frame subtraction method, the two frames subtraction and the three frames subtraction are used. The accuracy of the abstracting of the moving areas and the minimum of the cost of the memory and the CPU of the system are considered, and the optimization of the two faces was realized.In the mobile's locating, the centroid locating method is used, which caculates the geometric center of the moving area as the coordinate of the mobile. In the paper, the center of the circum-rectangle of the moving area is caculated. In the tracking of the mobile, there is the problem of tracking lag in the usual passive tracking methods because the mobile moves while the video turns. So the Kalman forecasting arithmetic is used, and the location of the mobile was forecasted in the next frame, and the tracking lag was eliminated.In drives system, the stepping motor of the pan-tilt is used, and the accuracy and the real time effect of the system controlling was improved.