计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (19): 28-31.

• 博士论坛 • 上一篇    下一篇

一种基于空间直方图的多特征目标跟踪算法

徐  琨1,2,杨小军2,郭兰英2   

  1. 1.陕西省道路交通智能检测与装备工程技术研究中心,西安 710064
    2.长安大学 信息工程学院,西安 710064
  • 出版日期:2012-07-01 发布日期:2012-06-27

Multi-feature object tracking algorithm based on spatial histogram

XU Kun1,2, YANG Xiaojun2, GUO Lanying2   

  1. 1.Shaanxi Province Road Traffic Intelligence Test and Equipment Engineering Technology Research Center, Xi’an 710064, China
    2.School of Information Engineering, Chang’an University, Xi’an 710064, China
  • Online:2012-07-01 Published:2012-06-27

摘要: 为提高跟踪算法对光照或背景的大幅度变化和车辆大范围运动的鲁棒性,提出了一种基于空间直方图的多特征目标跟踪算法。算法以自适应权值多特征乘性融合框架为基础,分别建立目标的颜色、边缘和纹理空间直方图,使用Mean Shift迭代,利用各特征空间概率分布图中目标与背景的BH系数,调整特征权值。该算法使跟踪不再过分依赖某一单一特征,实现了复杂背景下目标的准确跟踪。

关键词: 目标跟踪, 多特征, 空间直方图, 均值漂移

Abstract: To increase the robustness to the variation of illumination or background and the large overall motion of the object, a multi-feature tracking algorithm based on spatial histogram is presented. The proposed algorithm is based on the framework of adaptive weights production fusion. Three features spatial histogram of the object are established. The Mean Shift iterative solution is deduced. According to BH coefficient in the sub-feature spatial probabilistic distribution images, the weights are adjusted. The tracking results no longer depend on the single feature too much and the accurate tracking in complex background is realized.

Key words: object tracking, multi-feature, spatial histogram, mean-shift