计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (33): 191-193.DOI: 10.3778/j.issn.1002-8331.2008.33.058

• 工程与应用 • 上一篇    下一篇

基于高斯模板分层匹配的红外点目标检测算法

聂洪山1,2,陈晓飞3,4,赵 琪4   

  1. 1.中科院 软件所,北京 100190
    2.国防科技大学 ASIC研发中心,长沙 410073
    3.北京航空航天大学,北京 100083
    4.北京图形研究所,北京 100029
  • 收稿日期:2008-08-01 修回日期:2008-09-25 出版日期:2008-11-21 发布日期:2008-11-21
  • 通讯作者: 聂洪山

Multi-layer matching algorithm based on Gaussian model for small targets detection in infrared images

NIE Hong -shan1,2,CHEN Xiao-fei3,4,ZHAO Qi4   

  1. 1.Insitute of Software,Chinese Academy of Sciences,Beijing 100190,China
    2.ASIC R&D Center of NUDT,Changsha 410073,China
    3.Beihang University,Beijing 100083,China
    4.Beijing Institute of Graphics,Beijing 100029,China
  • Received:2008-08-01 Revised:2008-09-25 Online:2008-11-21 Published:2008-11-21
  • Contact: NIE Hong-shan

摘要: 介绍了高斯模板相关匹配红外小目标检测方法,指出了该方法的不足。提出了一种基于高斯模板的红外小目标分层匹配算法,这种方法避免了目标尺寸与模板尺寸失配情况下的匹配不精确问题,能够在不知道目标尺寸参数的情况下匹配到高斯形状的目标。实验结果表明这种方法很好地解决了低信噪比点目标检测问题。

关键词: 高斯模板, 点目标, 相关系数, 匹配滤波

Abstract: The relative matching targets detection method with gaussian model is mentioned,and the shortcomings of the method are pointed out.A multi-layer matching algorithm based on Gaussian model for IR small targets detection is introduced.With this algorithm,Gaussian targets can be detected without knowing their size parameters of X and Y dimensions.Experimental results show that this algorithm can effectively detect small targets in infrared image with low SNR.

Key words: Gaussian model, small targets, relative coefficient, relative matching