为了能够从单一SAR图像中提取出建筑物的三维信息,基于SAR成像原理提出了从图像中建筑物叠掩和阴影区域计算平面矩形屋顶尺寸以及高度的方法。针对SAR平面矩形屋顶建筑物仿真图像,利用灰度直方图信息定位建筑物二次散射区域,通过线灰度累加分割出叠掩和阴影区域,再采用恢复公式计算屋顶尺寸和高度。实验结果表明,对于具有明显几何形变的平面矩形屋顶建筑物单一SAR图像,此方法有效恢复屋顶尺寸和高度,提高了建筑物检测识别的准确度和精度。
In order to extract three-dimension information of buildings in single SAR images, a way to compute the roof dimension and height is proposed. Based on SAR imaging principle, this method uses the information of overlay and shadow areas of buildings in SAR images. In SAR simulation images including buildings with flat and rectangle roofs, double scattering area of buildings can be oriented by gray histogram. The overlay and shadow areas can be segmented by gray accumulation in lines, and the roof dimension and height can be computed by recovery formula. The experiment results prove that this way can recover the roof dimension and height effectively in single SAR images, including buildings with flat and rectangle roofs which have geometric distortions. The accuracy of detecting and recognizing buildings in SAR images can be improved.