刘振涛, 王朝英, 刘卫群. 基于多特征的双模板自适应更新跟踪算法[J]. 应用光学, 2012, 33(1): 78-84.
引用本文: 刘振涛, 王朝英, 刘卫群. 基于多特征的双模板自适应更新跟踪算法[J]. 应用光学, 2012, 33(1): 78-84.
LIU Zhen-tao, WANG Zhao-ying, LIU Wei-qun. Tracking algorithm of dual template adaptive updating based on multiple features[J]. Journal of Applied Optics, 2012, 33(1): 78-84.
Citation: LIU Zhen-tao, WANG Zhao-ying, LIU Wei-qun. Tracking algorithm of dual template adaptive updating based on multiple features[J]. Journal of Applied Optics, 2012, 33(1): 78-84.

基于多特征的双模板自适应更新跟踪算法

Tracking algorithm of dual template adaptive updating based on multiple features

  • 摘要: 针对均值漂移算法中采用单一颜色特征以及缺乏必要模板更新方法的缺陷,提出一种基于多特征的双模板自适应更新目标跟踪算法。引入像素点邻域灰度均值差和分层空间信息加强目标特征的鉴别性,再通过对目标与背景区域双模板相似度系数的综合分析,准确地判断跟踪干扰因素的来源,并以当前帧目标区域的相似度系数为权值对目标模板进行加权更新,使得模板更新速度与其目标特征变化相适应的同时抑制模板过更新,较好地解决了模板更新时机和更新速度等问题。仿真结果表明,所提算法在不同跟踪场景下均具有较强的鲁棒性。

     

    Abstract: To solve the problem of single color feature and the lack of necessary template updating method in mean shift algorithm, an adaptive tracking algorithm of dual template adaptive updating based on multiple features is proposed. The difference of gray mean in the neighborhood of pixel point and the layered space information are introduced to enhance feature differentiation. By analyzing the similarity coefficient of dual template in target and background region, the origin of disturbance factor is estimated truly, and the target template in target region of current frame is updated by similarity coefficient. The speed of template updating is adapted to the transformation of target feature, and at the same time excessive template updating is restrained, which solves the problem of the occasion and speed of template updating. The robustness of the proposed algorithm is demonstrated by simulations in different tracking scenarios.

     

/

返回文章
返回