论文部分内容阅读
通过遥感影像直接提取出建筑物是进行城市土地利用调查和土地执法检察的有效手段。文章利用高分辨率的SAR数据,提出了基于SVM和高分辨率SAR图像的建筑物自动提取方法。该方法利用SAR图像对建筑物信息反映强烈的特点,选择一系列建筑物和非建筑物像元作为训练样本,利用SVM方法构建每个像元的属于建筑物可信度,设定95%的可信度界定为建筑物像元,据此识别出建筑物的位置。以温江地区的COSMO-SkyMed数据进行试验表明,该方法建筑物识别的抽样精度达到93%以上,显示出高分辨率SAR图像在城市土地利用研究中的巨大潜力。
The direct extraction of buildings from remote sensing images is an effective means to carry out urban land use investigation and land enforcement prosecution. Based on high resolution SAR data, this paper proposes an automatic building extraction method based on SVM and high resolution SAR images. In this method, a series of buildings and non-building pixels are selected as training samples by using SAR images to reflect the strong information of buildings. The trustworthiness of each pixel belonging to buildings is constructed using SVM method, and 95% Credibility is defined as a building pixel, from which the location of the building is identified. Experiments with COSMO-SkyMed data in Wenjiang area show that the sampling accuracy of the method for building identification is above 93%, indicating the great potential of high-resolution SAR images in urban land use research.