基于决策融合和模型更新的红外目标跟踪方法  

Infrared Visual Tracking Via Decision Fusion and Model Updating

作　　者：李佳文 李建 彭程[1] 杨杰 LI Jia-wen;LI Jian;PENG Cheng;YANG Jie(Institute of Image Processing&Pattern Recognition of Shanghai Jiaotong University,Shanghai 200240,China;Shanghai Aerospace Control Technology Institute,Shanghai 200240,China;Infrared Detection Technology Research&Development Center of CASC,Shanghai 200240,China)

机构地区：[1]上海交通大学图像处理与模式识别研究所,上海200240 [2]上海航天控制技术研究所,上海200240 [3]中国航天科技集团公司红外探测技术研发中心,上海200240

出　　处：《计算机仿真》2021年第10期111-115,共5页Computer Simulation

摘　　要：红外图像中的目标跟踪技术在导弹制导等领域有着广泛的应用,是现代化军事实力不可缺少的一部分。传统方法主要构建目标的轮廓纹理模型,但对于模糊目标、快速移动目标或存在外界遮挡等情况并不鲁棒。提出一种基于动态决策融合和自适应模型更新机制实现的跟踪算法,利用轮廓纹理和红外能量分布信息共同描述目标,在不同情况下动态计算各支路权重以及模型的更新因子,从而充分利用了不同模型在各种情况下的优势,并避免错误的目标表观信息更新至模型。实验表明,提出的方法有效提高了算法在目标快速移动、形态模糊和外界遮挡等情况下的鲁棒性。Infrared visual tracking has a wide range of applications in many fields such as missile guidance. It is one of the most important parts of modern military strength. Traditional tracking approaches mainly construct the contour and texture models of the target, which is not robust enough to fuzzy target, fast-moving target, occlusion and other situations. This paper proposes a tracking algorithm based on the dynamic decision fusion strategy and the adaptive model updating scheme. The contour texture and infrared energy distribution information were used to describe the target together. The weights of each branch and the update factor of the model were dynamically calculated under different conditions so that different models were fully utilized under different situations. In addition, the proposed method could prevent the model from the wrong target apparent information. Experimental results demonstrate that the proposed approach is significantly robust to fast-moving, occlusion and fuzzy targets.

关 键 词：目标跟踪 红外图像 决策融合 自适应算法 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]