基于运动与颜色直方图的粒子滤波目标跟踪  

Object tracking based on motion and color histogram with particle filter

作　　者：吴雪刚[1] 

机构地区：[1]长江师范学院计算机科学系,重庆408003

出　　处：《计算机工程与设计》2008年第15期3968-3971,共4页Computer Engineering and Design

摘　　要：提出了基于运动目标历史速度和历史运动曲线的改进的粒子滤波器设计方案。为了解决运动中的遮挡问题,在粒子评价过程中引入"运动动量因子"来保持在高速运动下原来方向粒子的健壮性,提高了粒子跟踪的准确度。相应的为运动目标建立颜色模型,进一步增加了跟踪的准确性。实验结果表明,跟踪的效果优于基本粒子滤波器。By analyzing the velocity of motion object, the basic particle filter based the velocity and curve of history images is improved. To solve the occlusion problem, adding the motion inertia factor to the process keeps haleness of the particles and improves the tracking veracity. And, modeling color of the motion object improves the veracity of tracking motion object. The experiment shows the result is better than basic particle filter.

关 键 词：目标跟踪 卡尔曼滤波 粒子滤波 粒子退化 速度 遮挡 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]