利用纹理特征提取城市用地信息方法探索

就利用灰度共生矩阵纹理特征提取城市用地信息做了初步探索。计算灰度共生矩阵四个纹理特征量,选择建筑用地与其它地类的纹理特征统计量差别较大的特征,用于提取建筑用地信息。通过计算选择了对比度纹理特征,对该特征图像进行分类、密度分割及后处理,得到城市用地信息。通过精度评定证明了纹理特征用于分类可以提高分类的精度,并能提高土地利用动态监测的自动化程度。

Probe of texture features in city land use information extraction

Dynamic change detection is a central task for land use monitoring by remote sensing.The main methods currently used in land-use dynamic detection are computer human interaction or spectral classification.However,the former one is tremendous and thus demanding to operators and the latter is not accurate.The deficiencies mentioned above limit the popularity of remote sensing land-use dynamic detection.On the other hand,with the improvement of satellite resolution,remote sensing images have plenty of spatial information and how to use texture information to improve the accuracy and efficiency of land-use dynamic detection becomes important.This paper brings forward a new method of information extraction which uses texture features to extract city land-use information.According to statistics and analysis,this paper chooses contrast texture feature and uses texture classification to extract information about city land use.The experimental results illustrate its high performance in land-use dynamic detection with remotes sensing.