基于纹理特征的SPOT-5影像马尾松毛虫害信息提取  

SPOT-5 Image Texture Analysis Based Dendrolimus punctatus Damage Information Collection

作　　者：亓兴兰[1,2] 刘健[1,3] 胡宗庆[2] 余坤勇[1] 雷泽兴[2] 

机构地区：[1]福建农林大学林学院,福建福州350002 [2]福建林业职业技术学院,福建南平353000 [3]三明学院,福建三明365000

出　　处：《西南林业大学学报（自然科学）》2012年第1期46-50,111,共6页Journal of Southwest Forestry University：Natural Sciences

基　　金：国家自然科学基金项目(30871965)资助;福建省教育厅科技项目(闽教科〔2011〕33号)资助;南平市科技局项目(N2010Z19)资助

摘　　要：以福建沙县为研究区,以SPOT-5影像为数据源,采用灰度共生矩阵方法提取健康林分与受害林分的纹理特征,构建最佳纹理量,分别采用像元统计和面向对象的方法进行虫害信息提取,结果精度分别为72.00%、74.75%。研究结果证明了利用遥感影像纹理特征进行马尾松毛虫害监测的可行性,为利用融合影像光谱信息与纹理信息进行虫害信息提取研究提供了实例支撑和技术参考,同时面向对象的方法优于传统的基于像元统计的分类方法,精度稍高,"椒盐现象"也有所改善。The high-resolution image is extensively used to monitor forest diseases and insect pests with the development of remote sensing techmology.Taking Shaxian County of Fujian Province as the study area.and the SPOT-5 image as data source,the texture feature of both healthy and infested Pinus massoniana stands by Dendrolimus punctatus were extracted and the best texture were constructed with Gray-level Co-occurrence Matrix method.The pest information was derived respectively by pixel statistics and object-oriented methods,and the individual precision was individually 72.00% and 74.75%.The results showed that it was practicable to monitor the infestation of Dendrolimus punctatus with texture feature of remote sensing image,which provided the relevant research with case study support and technical reference.It was also showed that the object-oriented method with higher precision and better image quality was superior to the traditional pixel statistics classification method.

关 键 词：马尾松毛虫害 信息提取 纹理特征 SPOT-5影像 地面调查 

分 类 号：S76[农业科学—森林保护学]