针对建筑物特征信息提取的效率和精度问题,提出了一种基于点云切片和最小包围矩形的建筑物特征信息提取算法。该算法先对点云进行去噪和切片处理,然后利用每个切片中相邻点之间的水平角和竖直角差值自动快速提取建筑物整体和门窗的轮廓点,之后对轮廓点进行滤波、分类和排序,最后采用最小包围矩形提取规则建筑的整体尺寸和门窗尺寸信息。通过三组实验,将提取出的建筑物整体和门窗尺寸与实际尺寸进行比较,结果表明:所提算法对建筑物门窗和整体尺寸的提取准确度在3 cm以内,总体精度可达97.4%以上,对160万左右建筑物点云数据的提取总时间在8 s以内,证明了所提算法的有效性。

In this study, we propose an algorithm for extracting the building feature information based on point cloud slices and minimum bounding rectangles to enhance the efficiency and accuracy of building feature information extraction. First, the algorithm denoised and sliced the point cloud horizontally and vertically. Then, the horizontal and vertical angle differences with respect to the adjacent points in each slice were used to automatically and quickly extract the contour points of the entire building, doors, and windows. Subsequently, the contour points were filtered, classified, and sorted. Finally, the overall sizes of the regular building and the information of windows and doors were extracted using the minimum bounding rectangle. In this study, three experiments were conducted to compare the extracted sizes of the entire building, windows, and doors with their actual sizes. The results denote that the information extraction accuracy with respect to windows, doors, and the entire building is within 3 cm, the overall accuracy is greater than 97.4%, and the time required to extract the information of approximately 1.6 million building point cloud data is less than 8 s, proving the effectiveness of the proposed algorithm.