基于模型互更新的多模图像融合跟踪算法

Multimodal Image Fusion Tracking Algorithm Based on CoUpdate

针对复杂环境下引起的目标失跟问题,提出了一种基于模型互更新的可见光与红外图像融合跟踪算法。基于把视觉跟踪问题视为"中心-周围"分类的思想,首先从可见光与红外图像中分别提取目标及周围像素点的特征,然后采用Boosting算法训练得到跟踪模型。基于分类结果计算像素点的置信度,采用决策级融合方法得到似然图像,通过均值漂移算法估计目标位置。最后在Co-Training框架下结合目标跟踪结果进行模型的互更新。实验结果表明,该算法提高了跟踪的鲁棒性,有效利用了多模图像的信息。

A visible and infrared image fusion tracking algorithm based on Co-Update is proposed in this paper to solve the problem of complex environment caused by the target tracking lost. Treating the visual tracking problem as the thought of ＂Center-Around＂ classification, the proposed algorithm extracts the pixels characteristics of the target and surrounding from the visible and infrared image first, and then obtains tracking model by the Boosting algorithm training. The confidence coefficient of pixels is calculated based on the classification results, decision level fusion method is adopted to get likelihood image, and the Meanshift algorithm is used to estimate target position. Finally in the Co-Training framework the target tracking results are combined with to Co-Update the tracking model. The experimental results show that the proposed algorithm can improve the robustness of tracking and use the multimode image information effectively.