基于二阶直方图的改进型均值偏移跟踪算法  

Improved Mean-shift Tracking Algorithm Based on Second Order Histogram

作　　者：陈善静[1,2] 杨华[1,2] 曾凯[1,2] 杜石明[3] 张红[4] 

机构地区：[1]脉冲功率激光技术国家重点实验室,安徽合肥230037 [2]电子工程学院,安徽省红外与低温等离子体重点实验室,安徽合肥230037 [3]电子工程学院安徽省红外与低温等离子体重点实验室,安徽合肥230037 [4]电子工程学院物理教研室,安徽合肥230037

出　　处：《探测与控制学报》2010年第4期73-77,共5页Journal of Detection & Control

基　　金：国家“八六三”计划项目资助(2006AAJ103)

摘　　要：针对在均值偏移跟踪算法中颜色直方图忽略了颜色空间分布信息的问题,提出了基于二阶直方图改进型均值偏移跟踪算法。使用由颜色空间均值向量和颜色空间协方差矩阵所决定的wu对颜色直方图进行了改进,推导出基于二阶直方图的均值偏移向量。仿真实验表明:基于二阶直方图的改进型均值偏移跟踪算法达到了充分利用颜色空间分布信息的目的,提升了均值偏移跟踪算法的抗局部遮挡能力。The color histogram ignores the color spatial distribution information in mean-shift tracking algorithm.An improved mean-shift tracking algorithm based on second-order histogram was proposed.The color histogram was improved by wu,which was determined by color spatial mean vector and color spatial covariance matrix.Then mean-shift vector based on the second order histogram was derived.Simulation results showed that the improved mean-shift tracking algorithm based on second order histogram could make fully use of the color spatial distribution information,and the anti-partial occlusion capacity of mean-shift tracking algorithm was upgraded.

关 键 词：均值偏移(mean-shift) 二阶直方图 颜色空间分布 跟踪算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]