基于图像多尺度熵的红外图像匹配跟踪算法  

Infrared image matching tracking algorithm based on image multi-scale entropy

作　　者：刘兴淼[1] 王仕成[1] 赵静[1] 

机构地区：[1]第二炮兵工程学院科研部,西安710025

出　　处：《控制与决策》2011年第5期768-772,共5页Control and Decision

基　　金：国家技改重点项目子课题(WX07233)

摘　　要：在对图像熵进行分析的基础上,引入图像多尺度熵的概念,定义了图像的多尺度熵及多尺度熵矢量,提出了一种基于区域的匹配跟踪算法—–基于图像多尺度熵的红外图像匹配跟踪算法.首先计算图像的多尺度熵,得到图像多尺度熵矢量;然后利用多尺度熵矢量间的绝对距离(AD)进行匹配跟踪.实验表明,该算法不仅具有稳定、精确的匹配跟踪性能,而且能在目标发生旋转时,较好地匹配跟踪目标,并具有良好的抗几何失真能力.After analyzing the entropy of the image,the conception of image multi-scale entropy is introduced,and image multi-scale entropy and the vector of image multi-scale entropy are defined.A region matching tracking method-infrared image matching tracking algorithm based on image multi-scale entropy is present.Firstly,the image multi-scale entropy and the vector of the image multi-scale entropy are calculated.Then the absolute distance（AD） of the vector of image multi-scale entropy is applied to matching and tracking.The experimental results show that the proposed algorithm has the robustness matching tracking properties and is robust to the problems of the rotation and geometry distortion of targets.

关 键 词：图像匹配 多尺度熵 图像跟踪 红外图像 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]