基于纹理的风蚀水蚀过渡区TM影像土地利用信息提取  

Extracting land use information of TM image in wind-water erosion interlaced region based on Texture

作　　者：邓锟[1,2] 常庆瑞[1] 蔚霖[1] 王德彩[1] 

机构地区：[1]西北农林科技大学资源环境学院,陕西杨凌712100 [2]中煤航测遥感局遥感应用研究院,陕西西安710054

出　　处：《西北农林科技大学学报（自然科学版）》2009年第1期91-97,104,共8页Journal of Northwest A&F University(Natural Science Edition)

基　　金：国家自然科学基金项目(30571527);国家科技支撑计划项目(2006BAD09B0603)

摘　　要：【目的】探索在风蚀水蚀过渡区,利用TM遥感影像提取土地利用信息的方法。【方法】以榆林市西北部风蚀水蚀过渡区为对象,应用主成分变换与基于灰度共生矩阵纹理特征相结合的方法对TM影像进行分析,选取均值作为分类指标,采用分层分类法进行信息提取。【结果】不同土地利用类型之间纹理特征参数差异明显,荒漠化土地的均值、方差值较高,灰度分布不均匀,相邻像元反差明显,内部变化剧烈;水体和灌溉耕地的均值、方差值很低,灰度值均匀,内部变化小;林地和旱耕地信息介于上述两类土地之间。【结论】与传统分类方法相比较,纹理分析方法可以获得更高的分类精度。[Objective] A method for deriving land use information of TM image in the wind-water erosion interlaced region was discussed. [Method] This paper employed principal component analysis combined with Gray Level Co-occurrence Matrix texture analysis to abstract the land use information from the TM image in the wind-water erosion interlaced region （Yulin）. The classification indexes were selected according to mean and the land use information was abstracted by hierarchical classification. [Result] The results showed that there were significant differences among parameters of texture feature of the different land utilization. In the desertification, the mean and variance were comparatively high. The gray level was asymmetry, the contrasts between neighboring pixels were significant; while in water and irrigable land, the mean, variance was a little bit low, the grey level was symmetry;the mean, variance and gray level in the forest and dry land was between the above（desertification and water and irrigable land）. [Conclusion] Compared with the traditional classification method, the classification based on the texture analysis could acquire higher classifying precision.

关 键 词：灰度共生矩阵 主成分变换 风蚀水蚀过渡区 榆林市 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]