基于CBERS-02B的矿区地物信息的提取

Features information extraction of the mining area based on CBERS-02B

应用面向对象分类方法,对山西省平朔露天煤矿进行信息提取。以中巴资源卫星CBERS-02B卫星遥感影像为数据源,利用面向对象分类方法进行多尺度分割并建立分割等级,通过对不同分割尺度的分割结果进行比较,获得最终分割尺度,并结合具体地物的光谱、空间等特征,采用模糊分类中的隶属度函数分类方法,最终将地物分为植被、道路、矿区建筑、煤堆、开采面、废石堆六类。最后使用误差矩阵对分类结果进行精度评价,其总体分类精度达到了88.63%,Kappa系数为0.89。

Object-oriented classification method is used to extract information from the Pingshuo surface coal mine in Shanxi Province. China-BraziI Earth Resources Satellite（CBERS-02B）is used as data sources.The multi-scale segmentation and partitions level are created by object-oriented classification method.Through comparison of different segmentation results, the final segmentation scales are obtained.And spectral and spatial characteristics of the specific surface features are used to classify, the image into vegetation, roads, mine construction, coal pile, mining face, waste rock dumps through the fuzzy classification of the membership function classification method.In the end, The classification result is evaluated by the error matrix, the result shows that the overall classification accuracy reaches 88.03%, and Kappa coefficient is 0.88.