基于多特征分层匹配算法的颗粒目标跟踪

Research on Grains object tracking Based on Correlation Matching

在大量的相似小颗粒物体聚集图像中,目标颗粒与环境颗粒难以区分。针对这个问题,本文采用颗粒本身的灰度值与不变矩作为颗粒的特征,实现颗粒目标在平移、旋转和对比度发生变换时的准确匹配。采用规则预测的方法缩小匹配搜索范围。实验结果表明本文提出的颗粒目标跟踪算法可获得令人满意的跟踪效果。

To the question that difficult partition of the object grains and the background grains when grains is assembled largely, the paper adopt ash-degree and invariant moments as character of the grains, satisfying the matching when the object grains take place to parallel move, circumrotate or contrast changed, adopt a forecast algorithm based on orderliness to reduce the searching scope of matching, and in the process of tracking, the template is corrected by matching result in each frame. The last experiment testifies that this paper put forward the tracking algorithm′s accuracy achieves famous tracking effect.