欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2010, Vol. 27 ›› Issue (3): 370-375.DOI: 10.7523/j.issn.2095-6134.2010.3.010

• 论文 • 上一篇    下一篇

一种基于协方差矩阵的自动目标检测方法

宁忠磊1,2,3, 王宏琦1,2, 张正1,2,3   

  1. 1. 中国科学院电子学研究所,北京 100190;
    2. 中国科学院空间信息处理与应用系统技术重点实验室,北京 100190;
    3. 中国科学院研究生院,北京 100049
  • 收稿日期:2009-10-10 修回日期:2010-01-18 发布日期:2010-05-15
  • 基金资助:

    国家自然科学基金(40701110)资助 

An automatic object detection method based on covariance matrix

NING Zhong-Lei1,2,3, WANG Hong-Qi1,2, ZHANG Zheng1,2,3   

  1. 1. Institute of Electronics, Chinese Academy of Sciences, Beijing 100190, China;
    2. Key Laboratory of Technology in Geo-Spatial Information Processing and Application System, Chinese Academy of Sciences, Beijing 100190, China;
    3. Graduate University, Chinese Academy of Sciences, Beijing 100049, China
  • Received:2009-10-10 Revised:2010-01-18 Published:2010-05-15

摘要:

为了将协方差矩阵算法应用于自动目标检测,提出了特征相似度和协方差矩阵相似度.特征相似度是目标特征的相似程度,协方差矩阵相似度融合各个特征相似度.另外,鉴于特征具有不同的有效性和重要性,提出了最小特征相似度.最小相似度可以用于剔除基本无效的特征.通过实验证明,本方法能有效地将协方差矩阵算法应用于自动目标检测,具有较高的准确率.

关键词: 协方差矩阵, 自动目标检测, 特征融合

Abstract:

In order to apply the covariance matrix algorithm to automatic target detection we present feature similarity and covariance matrix similarity. Feature similarity is the similarity of the target feature. Covariance matrix similarity integrates all the feature similarities. In addition, because features are different in validity and importance, we raise minimized feature similarity. Minimized feature similarity can be used to get rid of basically ineffective features. Experiments show that with this method one can effectively apply the covariance matrix algorithm to automatic target detection with high detection rate and low false alarm rate.

Key words: covariance matrix, automatic target detection, feature fusion

中图分类号: