利用复合光谱纹理特征进行城市边缘区不透水层提取

发布时间：2018-05-15 07:23

本文选题：不透水层 + 纹理　；参考：《测绘通报》2016年11期

【摘要】：城市边缘区是城市化活跃的地区,在遥感影像上呈现出错综复杂、异质强的特征,对该地区不透水层的遥感提取带来一定的困难。本文引用扩展支撑向量机(extended support vector machines,ESVM)的软硬分类方法,结合光谱和纹理特征变化,对正在发生快速土地覆盖变化的城市边缘区的不透水层进行了提取。本文提出的方法将纹理变化作为有效信息表示城市边缘区的变异,并结合软硬分类方法的特征进行了不透水层纯净、混合像元的识别。研究区城市边缘区的试验结果表明,将纹理特征作为描述不透水层的空间特征指数能够将纯净、混合像元的识别效率提高10%,整体分类精度提高1%~5%,优于传统的软分类、硬分类方法。
[Abstract]:The urban fringe is an active area of urbanization, which presents complicated and heterogeneous features in remote sensing images, which brings some difficulties to remote sensing extraction of impermeable layers in this area. In this paper, the soft and hard classification method of extended support vector machine (extended support vector machine) is used to extract the impermeable layer in the urban fringe where rapid land cover changes are taking place in combination with the changes of spectral and texture features. The method proposed in this paper takes the texture change as the effective information to represent the variation of the urban fringe, and combines the features of the soft and hard classification method to identify the pure impervious layer and mixed pixel. The experimental results in the urban fringe of the study area show that using the texture feature as the spatial feature index to describe the impermeable layer can improve the recognition efficiency of the mixed pixel by 10 percent and the overall classification accuracy by 1 / 5, which is superior to the traditional soft classification. Hard classification method
【作者单位】：北京工业职业技术学院;北京师范大学资源学院;
【基金】：国家自然科学基金(41301444) 北京高等学校“青年英才计划” 北京工业职业技术学院一般课题(bgzyky201605)
【分类号】：P237

【相似文献】