Research And Implementation Of The Shot-put Measuring System Based On Monocular Vision

In recent years, with the rapid development of image processing technology, the vision measurement technology research has made great progress. It also has been widely used in many areas, such as industrial parts detection, vehicle range and robot vision. Using this technique in modern sports can achieve a high degree of measurement automation, reduce the artificial error, avoid the black box operation, and ensure the fair and equity in the competition or test. This paper focuses on the application of vision measurement in the shot-put, and then designs a performance measurement system based on monocular vision. It has an important theoretical reference value for the promotion of vision measurement technology in the modern sports.Although most shot-put and tests are now using a laser rangefinder to measure results, there are still some measurement problems such as low efficiency and artificial factor influence. In practical measurement, the laser rangefinder should benchmark vertically and artificially aligned. However, vision measuring system designed in this paper just needs a target placed on the landing shot, reducing measuring time and greatly improving the efficiency of measurement. Future works includes incorporation of a motion tracking module in the system to acquire shot-put flying. The path analysis can realize automatic recognition of shot point, fully achieving realization of automatic measurement system.The main work is as follows:(1) Regarding the shot-point target recognition problem, a fast target recognition method based on background subtraction and multi-feature fusion is proposed. It uses the background subtraction method to filter out a large number of non-target information, and then uses the color and shape feature to recognize the target and distinguish them from other objects. This method can not only identify the target accurately, but also satisfies the system real-time requirements.(2) Based on the plane characteristics and application demands of shot-put field, the system adopts the corresponding points calibration method to calibrate camera. In order to improve the calibration precision of the camera, this paper proposes a homography matrix evaluation model which uses the least square method based on the distance between any two points. The coordinate of known points is not necessary, just the distance of any two points, easy to measure, and the error value is relatively small. The real coordinates of the shot-point are calculated by the optimal homograph matrix, and then the throwing distance is obtained. Though testing the system on the field, the actual measurement results conform to the requirements of measurement errorThe vision measurement system has simple structure, low cost and high feasibility. Moreover, there is still much room for improvement in the precision of the system. So the model has a good application and promotion prospects.