基于双直方图自适应选择的目标跟踪算法*

针对传统的均值漂移算法，加入了梯度方向直方图及其与颜色直方图的自适应选择，提高了均值漂移算法在复杂场景中目标跟踪的鲁棒性。传统的均值漂移算法往往选择固定的一个颜色直方图对目标进行跟踪，当目标自身或者跟踪场景发生变化时，容易跟踪失败。通过分析被跟踪目标在当前场景中与目标模板在颜色以及梯度方向上的相似性并设定阈值，从而选择并使用当前有效的目标特征，实现复杂变化场景下的目标跟踪。一系列不同场景下的运动目标跟踪实验，证实了该算法的可靠性。

Object tracking algorithm based on self-adaptive selection of double histograms

This paper improved the traditional mean shift tracking algorithm with self-adaptive selection of both color histograms and gradient-oriented histograms, thus strengthened the robustness of novel tracking algorithm in complicated circumstance. Since the traditional mean shift tracking algorithms were usually based on one fixed color histogram, it was prone to fail when used to track changeable objects or used in changeable circumstance. In the proposed algorithm, analyzed the color and gradient orient similarities between object in present circumstance and object templates and then set a threshold so that the most effective features to present object tracking were selected, thus realizing object tracking in complicated dynamic circumstance. The reliability of the improved algorithm has been verified in serial experimental results of moving object tracking in different circumstance.