论文部分内容阅读
本文提出了一种基于全色波段航空影像和激光雷达数据的建筑物检测方法。如何从激光点云数据中提取出建筑物激光脚点,是建筑物三维重建和轮廓提取的难点问题之一。植被密集区域以及与建筑物紧密相邻的树木的激光点很难与建筑物激光点区分开。本文利用支持向量机对单个激光点的特征进行两分类,特征向量包括激光点的高程、高程变化信息以及与激光点配准的影像光谱信息。实验表明,基于支持向量机的点态分类算法能够有效提取建筑物激光脚点,影像光谱信息能明显提高分类精度。
This paper presents a method of building detection based on panchromatic aerial imagery and lidar data. How to extract the laser footing from the laser point cloud data is one of the difficult problems in 3D reconstruction and contour extraction of buildings. Laser spots in densely populated areas and trees that are immediately adjacent to buildings are difficult to distinguish from building laser spots. In this paper, we use support vector machine to classify the characteristics of a single laser point. The feature vector includes the laser point elevation, elevation change information and image spectral information registered with the laser point. Experiments show that the point state classification algorithm based on support vector machines can effectively extract the laser foot and image spectral information of buildings can obviously improve the classification accuracy.