The Hawkeye system can track and record the trajectory of the ball on the field, and plays an important role in assisting the referee to decide the boundary ball. However, due to the problems of small target, fast flight speed, light changes, advertising characters and other moving targets in the tennis extraction process, the accuracy of tennis feature parameter extraction needs to be further improved. In order to solve these problems, an accurate extraction algorithm of moving target feature parameters in Hawkeye system was put forward. The algorithm made use of the characteristics of good color consistency of tennis, obtains the range of tennis tone according to color training, the tennis image was segmented in the HSV color space to reduce the interference of background image, the fusion algorithm of inter-frame difference and mixed Gaussian model was used to further segment the tennis image to get the moving target. Sobel operator was used to extract the edge information of the segmented binary image to reduce the amount of calculation of subsequent target circle detection and improve the efficiency. The precise extraction of tennis and radius parameters was accomplished by using Hough circle transform to detect contour edge information. The experimental results show that the feature parameter extraction algorithm has high fitting accuracy and small relative deviation, and has a certain practical application value.