空间相关背景加权直方图均值漂移目标跟踪

针对传统的基于直方图的均值漂移跟踪算法, 当目标出现于复杂跟踪条件时, 因无法显著区分颜色相近的目标或背景而导致跟踪失败的问题, 提出一种基于空间相关背景加权直方图的均值漂移跟踪算法, 并给出了推导过程. 实验结果表明, 该算法通过在目标模型中减少背景显著特征来增强目标定位, 能在遮挡、光照变化等复杂情况下实现有效的目标跟踪, 改善了传统均值漂移算法中模板不能及时更新等方面的局限性.

Mean shift target tracking with spatiogram corrected background-weighted histogram

For the traditional mean shift tracking algorithm based on histogram, tracking is failing when the target appears under the complicated tracking conditions. It can’t significantly distinguish color-similar targets and backgrounds. A kind of mean shift tracking algorithm based on spatiogram corrected background weighted histogram is proposed, and its detailed derivation is given. Targeting positioning is enhaced by reducing background prominent feature in target models. Experiment results show that the algorithm can effectively track the targets on complicated situations of occlusion and illumination variations, and can improve the limitations that traditional mean shift template can’t be updated.