基于北京二号影像辅助nDSM的建筑物自动提取

针对建筑物提取方法缺乏泛化性的问题,本文提出了将nDSM、北京二号影像、NDVI、BAI的七通道图像相结合作为数据源的提取方法。采用随机森林、梯度提升机、支持向量机、BP神经网络分类器对建筑物进行提取获取最佳分类器模型,并运用二值化与开闭运算,以建筑物面积与最小外接矩形面积的比值为阈值,对建筑物分别进行最小外接矩形、DP算法拟合,优化建筑物提取结果。试验结果表明,梯度提升机(GBDT)较其他分类模型在不同场景下综合效果较好,F-score精度更高。

Construction of building automatic extraction process based on image-aided nDSM of BJ-2

Aiming at the problem of lack of generalization of the building method of extraction,seven-channel images of nDSM,BJ-2 image,NDVI,and BAI are combined as the data source extraction method in this paper.Random forest,gradient boosting machine,support vector machine,BP neural network classifiers are applied to extract buildings to obtain the best classifier model;Binarization,opening and closing operations are applied,using the ratio of the area of the building to the area of the smallest enclosing rectangle is used as the threshold,and the smallest enclosing rectangle and DP algorithm are used to fit the buildings respectively to optimize the building extraction results.The experimental results show that the gradient booster (GBDT) has higher F-score accuracy when extracting buildings in different scenarios.