基于GLCM的纹理特征量在TM6热红外波段影像信息分析中的研究  

Textural Features Analysis Based on GLCM in TM Thermal Infrared Remotely Sensed Images

作　　者：宋杨[1] 万幼川[1] 陈鹏[1] 

机构地区：[1]武汉大学遥感信息工程学院,武汉430079

出　　处：《遥感信息》2006年第4期24-26,11,共4页Remote Sensing Information

基　　金：国家自然科学基金项目(资助号:60175022)

摘　　要：利用了基于灰度共生矩阵GLCM的纹理特征分析方法,对TM 6热红外波段影像进行目标信息的提取,以增强热辐射量大的亮温目标与背景图像的差异,从而实现对原始图像进行图像增强、实现目标信息提取的目的。本文的实验影像是从武汉市长江流域区域的TM影像的热红外波段影像上截取的16个子区域,对16个实验区域分别采用文中定义的6个纹理特征进行图像分析,原始图像的灰度级是256,考虑到计算量大,在处理过程中压缩原始的256灰度级到64灰度级。实验结果显示出较高的一致性,表明所定义的variance和correlation两个纹理特征对于在TM第6波段上,对于增强热红外影像上热辐射量大的目标影像与背景影像有显著的效果,该方法可以作为热红外影像处理的预处理,从一定程度上弥补了TM热红外波段影像分辨率不够高带来的限制。In this paper a computing six textural features based on co-occurrence matrix of infrared images is described.Usually the warmest object has the highest gray level values of infrared images,using the textural features should show which of them has the influence in highlighting a difference between the warm objects and their background.For that purpose the algorithm is developed and sixteen infrared images are tested.Due to computing intensity of the co-occurrence matrix the 256 gray-levels of infrared image were reduced to 64 gray-levels.Achieved results showed that variance and correlation contributed towards highlighting the warm objects and their background.

关 键 词：信息提取 纹理分析 灰度共生矩阵 

分 类 号：P237.9[天文地球—摄影测量与遥感]