| 基于粒子滤波/Mean Shift的改进跟踪算法 |
| |
| 引用本文： | 杜方芳,刘士荣,邱雪娜,刘斐. 基于粒子滤波/Mean Shift的改进跟踪算法[J]. 杭州电子科技大学学报, 2009, 29(6) |
| |
| 作者姓名： | 杜方芳 刘士荣 邱雪娜 刘斐 |
| |
| 作者单位： | 1. 杭州电子科技大学自动化研究所,浙江,杭州,310018
2. 华东理工大学自动化研究所,上海,200237 |
| |
| 基金项目： | 国家自然科学基金资助项目，浙江省科技计划资助项目，浙江省宁波市自然基金资助项目，浙江省教育厅科研资助项目 |
| |
| 摘 要： | 针对单一特征所带来的跟踪不稳定问题,该文提出一种基于纹理特征粒子滤波/Mean Shift的改进目标跟踪算法。该算法中建立一种选择反馈机制,首先对目标同时进行基于纹理信息的粒子滤波和基于颜色信息的Mean Shift两种算法的跟踪,然后对两种算法的跟踪结果进行比较,选择结果较好的输出,并把结果反馈到粒子滤波与Mean Shift中作为下一帧处理的初始值。实验结果表明,该方法克服了单一特征所带来的跟踪不稳定问题且具有较强的鲁棒性。
|
| 关 键 词： | 目标跟踪 粒子滤波 均值迁移算法 |
An Improved Object Tracking Algorithm Based on Particle Filter and Mean Shift |
| |
| DU Fang-fang,LIU Shi-rong,QiU Xue-na,LIU Fei. An Improved Object Tracking Algorithm Based on Particle Filter and Mean Shift[J]. Journal of Hangzhou Dianzi University, 2009, 29(6) |
| |
| Authors: | DU Fang-fang LIU Shi-rong QiU Xue-na LIU Fei |
| |
| Affiliation: | 1(1.Institute of Automation; Hangzhou Dianzi University; Hangzhou Zhejiang 310018; China; 2.Institute of Automation; East China University of Science and Technology; Shanghai 200237; China); |
| |
| Abstract: | Focusing on the problem of the instability brought about by the use of single feature,an improved object tracking algorithm was proposed in this paper based on particle filter with texture feature and mean shift with color feature.The feedback mechanism was modified in both particle filter and mean shift.Firstly,the object was tracked through particle filter with texture feature and mean shift with color feature simultaneously.After that,better result was selected by comparing the tracking results from two ... |
| |
| Keywords: | object tracking particle filter mean shift |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
