| Popis: |
With the rapid development of intelligent sensor technology and the application of scenarios, the video images of moving targets collected by intelligent terminal devices can be obtained in sports teaching scenarios, and then through preprocessing, target tracking and behavior analysis, the analysis report is finally generated to achieve the effect of sports training. However, in the process of processing, the edge information of the video image is damaged, the color deviation exists, and when tracking the moving target, the final result will be affected by the deformation and partial occlusion of the target. In view of the above problems, the original bilateral filtering algorithm is changed to the multi-scale Retinex algorithm in the preprocessing to ensure that the video image is clearer and more complete and can be effectively protected. At the same time, for the target tracking algorithm, the Camshift algorithm which takes the color histogram as the tracking target and records the distribution probability is used. So that the results will not be affected by the deformation and partial occlusion of the target. The final experimental results show that the improved pre-processing algorithm and target tracking algorithm have smaller errors in operation, better protection of the edge information of video images, reduced the time consumption of the target tracking unit to be detected, and also improved its anti-jamming stability. In the future development, the image processing algorithm should have a better operation path, further reduce the running time, improve the value of UCIQE and peak signal to noise ratio, and track multiple targets at the same time with the smallest error, capture the moment to reach the single frame level, and also become better for teaching. In the system design, information security performance is compared with vulnerability identification to evaluate the security level of physical education teaching system. |
