基于决策树方法的CBERS影像土地利用分类试验

Study on Land Use Classification of CBERS Image Based on Decision Tree Method

本文以陕西省西安市灞桥区与高陵县交界区域的中巴资源卫星数据为依据，在实地调查和光谱值分析的基础上建立了分类决策树，并进行了土地利用分类，得到了四种用地类型。从分类精度的定量评价来看，分类结果的精度较高，与土地利用的实际状况也比较吻合。决策树方法是一种有效的，灵活的，并且有着广泛应用前景的分类方法，设计更有效的分类器是提高决策树分类精度的关键。

In this paper, the author analyzes the CBERS remote sensing image of the boundary between Baqiao district and Gaoling county of Xi＇an in Shaanxi province. Based on field research and spectrum value analysis, a deci- sion- tree is created to classify the land use into four types. The quantitative evaluation of classification accuracy shows that the final accuracy is good and reflects the actual condition of land use well. The decision tree is an effective, flexible method to classify the land use in remote sensing image, and the method has a broad application pros- pect. The key to improve the accuracy of classification with decision tree is to design more effective classifier.