The Research On Moving Target Detection Based On Template Matching

The detection of a moving target is a important part of machine vision. Through the tracking work, a lot of human source will be reduced, it brings us convenience in our normal life and work, also it has show a bright future.This paper is aim at the tracking problem of aerocraft, the goal is to achieve the tracking goal in a certain background, basically adapt to the rotation and transformation. Many method have been used to do the track work, but the infrared and the ultrasonic technology have their weakness, they are easily infected by the barrier, and lost the target.Through the comparison of 4 methods in the target detection, we choose the Template Matching way to do the work because of its simpleness and easy to implement. In this paper, we use the USB camera to get the image frame dynamically, and then process the data we've just get by using the basic image processing methods, include a simple filter and the gray transformation. Because the original data has a huge quantity. In order to reduce the calculation. We have to transform the original data into the gray data. Based on the gray data we get, we use Roberts operator to calculate the contour data. How to calculate faster and nicer is the key to track the target.Template Matching work is very important. In this paper ,we use the contour data of the target to match. The traditional Template Matching way is very easy to be influenced by the background, etc, we have to do some improvement. So we add SSDA ,the Update, Transmutation and the Forecast ways, and optimize the matching principle and searching strategy. finally, combine these work into a intact matching arithmetic..Through the experiment result, the arithmetic is suitable to the tracking problem in a simple background, and can solve simple rotation and transformation problem, achieve the anticipate goal.