针对目标遮挡问题以及基于SSD和MCD跟踪方法的缺点,提出了一种新的基于SSD和MCD融合的跟踪方法,给出了SSD和MCD两种距离结合的目标相似度量的计算方法。新的方法既考虑了与模板像素距离较近点的个数同时也考虑了匹配区域的SSD距离。实验表明基于SSD和MCD融合的跟踪算法明显优于单独使用SSD和MCD的方法,特别是对目标部分遮挡的适应能力有了很大提高。
A new measurement of target similarity and tracking algorithm robust to partial occlusion based on fusion of SSD (Sum of Squared Difference) and MCD (Maximum Close Distance) was proposed. The number of pixels that its distance from template pixel is less than threshold and SSD of matching area are both taken into account in the new method. Experimental results show that the new algorithm is much better than the SSD-based or MCD-based method obviously,especially in the case of occlusion.