【摘要】 视频跟踪算法是计算机视觉实践课程中比较受关注的实验项目。针对突变情况下传统Mean Shift跟踪算法无法实时准确跟踪的问题,设计了基于模板更新和线性预估的Mean Shift跟踪算法创新实验项目。在模板更新策略下,引入背景模板,通过将原目标模板和背景模板与设定的阈值进行比较来对干扰因素进行判定,当干扰因素判定目标受到遮挡时,引入线性预估方程进行目标位置预测,有效解决目标在遮挡情况下跟踪丢失的问题。通过对测试视频的跟踪效果和性能进行对比分析,验证了算法在突变情况下相较于传统算法具有更好的抗干扰能力。以算法创新设计为核心,通过开放性创新实验项目的选题、设计、答辩、反馈的闭环实验过程,有效提高了学生算法创新设计能力。更多还原

【Abstract】 Video tracking algorithm is a more concerned experimental project in computer vision practical courses. Aiming at the problem that the traditional Mean Shift tracking algorithm cannot track accurately in real time under sudden changes, an innovative experimental project of Mean Shift tracking algorithm based on template update and linear estimation is designed. Under the template update strategy, the background template is introduced, and the interference factors are determined by comparing the original target template and the background template with the set threshold. When the interference factors determining the target is covered, the linear predictive equation is introduced to predict the target position, so as to solve the problem of target tracking loss under occlusion. Through comparative analysis of the tracking effect and performance of the test video, it is verified that the algorithm has better anti-interference ability than the traditional algorithm in the case of sudden changes. With algorithmic innovation design as the core, the closed-loop experimental process of open innovation projects, including topic selection, design, defense and feedback, has effectively enhanced students’ capabilities in algorithmic innovation design.更多还原