The precise analysis the details of the action and trajectory in the hurdlers technical videos can provide training deficiencies for the coaches.To realize the analysis of technical videos, we must firstly detect and track the athletes in the video. Based on the Opencv vision library of Microsoft Visual Studio 2008, we solve the problem of background shake by Surendra algorithm. Then the probability distribution of the target pixel to positioning the athletes is calculated using Kalman filter algorithm to predict its trajectory. Finally, we finish the recognition and detection of athletes by a combination of both methods.