基于颜色特征的多目标瞬间重合识别和跟踪

When moving objects are covered, it is easy for mean-shift algorithm to drop tracking. Then, this article advanced an multi-objects moment superposition identifying and tracking algorithm which was based color feature. Background deductiong was used to get the foreground objects. When ojects was superposed by the minute, Dynamic matching matrix was built by center-based color histogram to match objects and carry on mean-shift tracking. Experiments showed that after moving objects superposed and seperated by the minute, this algorithm could accurately track objects and could satisfy the need of real-time system.