Data Capture Method of Sports Video Trajectory Based on Dynamic Programming Algorithm

Stable and efficient video object trajectory tracking can fully reflect video information and has been widely used in different fields. As a novel study, this research proposes the data capture method of sports video trajectory based on dynamic programming algorithm. The innovation of the study are: 1. Improve the traditional monocular vision imaging model (NVI) from the imaging principle to reduce the imaging noise caused by lens angle and internal parameter errors, so as to obtain video information stably and efficiently. 2. Combined with the improved kernel correlation filter (KCF) algorithm and the innovative fusion of multi-channel HOG features, the target position is accurately tracked in the video frame sequence, thereby achieving trajectory tracking goals. 3. A multi-line position distance method is designed to effectively estimate the similarity of trajectories by calculating the weighted polygon area of the area enclosed by two trajectories. 4. In terms of trajectory optimization, a dynamic programming algorithm based on a clipping window is used, and grid division and clustering mechanisms are integrated to protect the information. The experiment is conducted under different video schemes and the results have shown the proposed algorithm can obtain the video trajectory effectively.