基于决策树模型的吉林西部居民地分布信息提取

Extraction to the Residential Distribution Information in West Jilin Based on Decision Tree Model

以吉林西部为研究区,将遥感影像的光谱特征和纹理特征作为分类依据,建立决策树模型,提取居民地信息,并选用手动提取的吉林西部居民地信息,利用混淆矩阵对其进行精度评价。结果表明,决策树易于综合光谱和纹理特征进行居民地信息提取,通过对比发现,利用决策树方法提取的居民地的精度明显高于传统的最大似然法,并总结出了利用该方法提取居民地信息的算法。

Targeted at the western part of Jilin Province,classified by the spectral signature and texture features of remote sensing image,a decision tree model to extract the information of residential evaluated by confusion matrix was set up.In addition,the calculation method to extract the information of residential was concluded.The research results indicated that compared with the traditional maximum likelihood method,decision tree model based on spectral signature and texture features was obviously more precise.