| 引用本文: | 侯晴宇,卞春江,逯力红,张伟,张晔.红外图像中快速小目标的均值移位跟踪[J].哈尔滨工业大学学报,2013,45(4):79.DOI:10.11918/j.issn.0367-6234.2013.04.015 |
| HOU Qingyu,BIAN Chunjiang,LU Lihong,ZHANG Wei,ZHANG Ye.Mean shift tracking for fast small target in IR imagery[J].Journal of Harbin Institute of Technology,2013,45(4):79.DOI:10.11918/j.issn.0367-6234.2013.04.015 |
|
| 本文已被：浏览 2627次 下载 1405次 |
码上扫一扫！ |
|
|
| 红外图像中快速小目标的均值移位跟踪 |
|
侯晴宇1,2, 卞春江1, 逯力红3, 张伟1, 张晔2
|
|
(1.哈尔滨工业大学 空间光学工程研究中心,150001 哈尔滨;2.哈尔滨工业大学 电子与信息工程学院,150001 哈尔滨;3.天津工业大学 理学院,300387 天津)
|
|
| 摘要: |
| 为了增强红外图像中快速小目标跟踪的稳健性,提出基于隶属度加权核直方图目标表征模型的改进均值移位跟踪算法．首先分析原始均值移位算法跟踪快速小目标的局部背景干扰问题,融合背景信息提出隶属度加权核直方图目标表征模型．该模型能够提高对于目标的表征能力,抑制局部背景干扰．然后,以Bhattacharyya系数作为相似性度量,在均值移位框架下推导出基于该模型的移位向量,能够有效实现目标的移位跟踪．移位过程中,目标隶属度大的灰度具有高移位权重,反之具有低移位权重．最后,应用模型更新方法克服局部背景的时变性,进一步提高目标跟踪的鲁棒性．实验结果表明,该算法对于红外图像中的快速小目标的具有稳健的跟踪性能． |
| 关键词: 目标跟踪 红外 快速小目标 均值移位 |
| DOI：10.11918/j.issn.0367-6234.2013.04.015 |
| 分类号: |
| 基金项目: |
|
| Mean shift tracking for fast small target in IR imagery |
|
HOU Qingyu1,2, BIAN Chunjiang1, LU Lihong3, ZHANG Wei1, ZHANG Ye2
|
|
(1.Research Center for Space Optical Engineering, Harbin Institute of Technology, 150001 Harbin, China;2.School of Electronics and Information Engineering, Harbin Institute of Technology, 150001 Harbin, China;3.School of Science,Tianjin Polytechnic University, 300387 Tianjin, China)
|
| Abstract: |
| To enhance the robustness of IR fast small target tracking, an improved mean shift tracking algorithm based on membership degree weighted kernel histogram target representing model is proposed. Firstly, the local background interference problem in tracking fast small target with original mean shift algorithm is analyzed and the membership degree weighted kernel histogram target representing model merged into background information is presented. This model is able to enhance the representing capability of target and suppress local background interference. Then, the shift vector of this model is deduced in the framework of mean shift by regarding Bhattacharyya coefficients as the similarity measure. The target shift tracking is achieved effectively according to target gray level of large membership degree with high shift weight, and vice versa with low shift weight. Finally, the local background time-varying is conquered by employing model updating method and the robustness of target tracking is further improved. The experimental result indicates that the algorithm can improve the shift weight of target pixel gray level and suppress background interference, thus the performance of tracking the IR fast small target is robust. |
| Key words: target tracking IR small target mean shift
|
