颜色与纹理自适应融合的粒子滤波跟踪算法  

A particle filter tracking algorithm based on adaptive fusion of color and texture

作　　者：朱文佳 金强[2] 余烨[2] 

机构地区：[1]安徽百诚慧通科技有限公司,安徽合肥230088 [2]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《合肥工业大学学报（自然科学版）》2017年第8期1053-1058,1105,共7页Journal of Hefei University of Technology：Natural Science

基　　金：安徽省科技攻关计划资助项目(1401b042009);安徽省高等学校自然科学研究资助项目(KJ2014ZD27)

摘　　要：针对传统粒子滤波跟踪算法中的目标模型单一、粒子退化等问题,文章提出了一种基于颜色特征与纹理特征自适应融合的粒子滤波跟踪算法,该方法选取颜色特征和纹理特征作为目标的视觉描述子,然后将2种特征的后验概率进行融合,并对目标模板进行自适应更新,进而估计出目标的状态。实验结果表明,该方法能够较好地处理背景光照变化明显、目标物体与背景颜色相近、遮挡、局部形变等干扰因素的影响,准确跟踪及定位运动目标。In order to solve the problems of the singularity of target model and particle degeneracy intraditional particle filter tracking algorithm, an improved particle filter tracking algorithm based on a-daptive fusion of color feature and texture feature is proposed. Firstly, color and texture features arechosen as visual descriptor of the object. Then the posterior probability of the two features is fused to-gether, and the object model is updated adaptively, and then the object state is estimated. Experimen-tal results show that the proposed method can accurately locate and track moving objects in case of ob-vious background illumination changes, similar object and background, occlusion, local deformationand other interfering factors existed in complex environment.

关 键 词：目标跟踪 粒子滤波 颜色特征 纹理特征 特征融合 

分 类 号：TP391.413[自动化与计算机技术—计算机应用技术]