基于SVM和线性光谱混合模型的城市不透水面丰度提取

Urban Impervious Surface Percentage Extracted in Hefei Based on Remote Sensing

不透水面丰度的提取在城市环境监测和规划研究中具有重要价值.基于2006年7月30日合肥市TM影像数据,运用支持向量机(SVM)监督分类与线性光谱混合模型(LSMM)相结合的方法提取研究区不透水面丰度值,并与单一运用线性光谱混合模型提取结果进行比较.研究结果表明:支持向量机监督分类和线性光谱混合模型相结合的方法RMSE为12.8%,提取精度高于线性光谱混合模型,该方法提高了城市不透水面丰度提取的精度.

Extracting impervious surface percentage using remote sensing technology is very important for monitoring urban environment and urban planning. A study was conducted to extract impervious surface percentage of Hefei based on TM image data of 30 July 2006 using the method of combining the support vector machine （SVM） to supervision classification and linear spectral mixture model （LSMM）. The extraction result was tested and compared with linear spectral mixture model method. The results of the study show that： the RMSE value of method of combining the support vector machine to supervision classification and linear spectral mixture model is 12.8%. The combined method was superior to the method of linear spectral mixture model, and the combined method imt）roved the accuracy of extraction of urban impervious surface percentage.