提出了一种新的完整地利用高分辨率光学图像和SAR图像进行城市矩形建筑物检测和高度估计的方法.该方法充分利用了两种数据的互补性,由三个顺序执行的步骤组成: 首先对两种数据分别进行建筑物检测,然后基于一种整合的准则对两种图像进行配准,最后利用一种“假设生成-仿真-匹配”的方法对建筑物高度进行估计.该方法对部分遮掩的建筑物也能取得好的效果.对两个不同场景的测试验证了该方法的有效性.

A new complete scheme for urban rectangle building detection and height estimation was proposed using one high-resolution optical image and one synthetic aperture radar (SAR) image. The scheme exploited adequately the complementarities of both data and consisted of a sequential processing chain: building boundary extraction from both data, registration from optical image to SAR image based on an integrated criterion and building height estimation based on a “hypothesis generation-simulation-matching” procedure. Especially, the proposed method performed well also for partially occluded buildings. The method was tested and validated over two different scenes.