一种重叠视域多摄像机环境下的运动目标匹配SHFSM算法  

Algorithm named SHFSM about multi-camera moving object matching in overlapping fields of view

作　　者：霍阔[1] 李长云[1] 李妍[2] 赵阳[1] 

机构地区：[1]湖南工业大学计算机与通信学院,湖南株洲412008 [2]深圳大学数学与计算科学学院,广东深圳518000

出　　处：《计算机应用研究》2013年第2期619-622,共4页Application Research of Computers

基　　金：国家技术创新基金资助项目(11C26214302856);国家自然科学基金资助项目(6077311);湖南工业大学研究生创新基金资助项目(CX1212)

摘　　要：针对重叠视域下多摄像机间目标匹配问题,提出了一种基于色度特征信息的色度特征区间集合匹配算法。通过分析目标颜色色度分量对光照强度的不变性,依据色度分量直方图定义了色度特征区间,并证明了其相关特性。该算法首先获取目标色度特征区间集合,并依据区间内特征信息像素的数量对特征区间进行排序,最后采用模糊匹配原理对排序的色度特征区间集合进行匹配。实验结果表明,算法匹配正确率可达95%,通过与经典算法对比体现了其准确率高、可靠性强的优点。The key problem of multi-camera object tracking is how to match the same object in overlapping fields of view be- tween different cameras. This paper introduced an algorithm named SHFSM which based on the information of hue feature. By analysing characteristic of hue component,it defined hue features section which named HFS, and proved its characteristic using hue component histogram. Firstly,this algorithm calculated the set of objects＇ HFS,then sorted the set based on the number of characteristic pixels distributed in HFS. Finally,it matched the set of object sorted by using fuzzy similar matching principle. Results obtained from experiments demonstrate that the algorithm achieves a correct matching rate of 95%. In contrast with classical algorithm, it has high accuracy and good reliability.

关 键 词：重叠视域 目标匹配 色度特征区间集合匹配 模糊匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]