Research On Automatic Stake Test System Based On Machine Vision

Machine vision is composed of many subjects. It is not only a scientific subject but an engineering discipline. Image processing technique and machine vision technique are adopted to discuss the method how to apply the machine technique theory to Automatic Stake Test System, which is valuable to the research and application of compute vision technique.In this paper, the object of research is the Automatic Stake Test System. How to extract and recognize the moving object through image processing technique and machine technique in any background to compute the distance of the object and track automatically are the emphases of research in this paper.1. A improved HIS image difference model based on sequences image difference and second image difference is presented. A method of self-adaptive division and elimination of the object shadow to extract the moving objects without background knowledge is adopted.2. The feature analysis and compute of moving objects are defined to feature match and recognize the area of objects, meanwhile the distance of the object barycenter is computed.3. The moving objects state forecast and two steps increment tracking algorithm based on Kalman filter can drive the mobile robot to track the moving objects quickly and smoothly4. Hough transform is adopted to obtain the information of the line in the area of test. The result of test can be gotten through comparing the position of line and the frame of car.A whole frame of system is designed in Windows environment. Experimental results show that the moving objects can be extracted and recognized through the arithmetic, and it is more stable an quickly to track the object. The justice and nicety of the test will be improved.