基于三维激光扫描数据的窗户边界提取方法

Window Boundary Extraction Method Based on 3D Laser Scanning Data

针对现有窗户提取算法中鲁棒性不高,边界规则化过程复杂的问题,该文提出了一种新的基于格网的窗户边界提取方法。首先采用PCA算法计算局部邻域内三个特征值,依据三个特征值的关系可完整提取建筑墙面点云;其次对墙面点云划分格网并二值化,粗提取窗户边界;然后基于粗提取的窗户边界格网统计分析,精确提取出窗户边界并二次建立格网;最后合并同一窗户的子窗户区域完成窗户边界的提取。经实例分析验证了该方法提取窗户边界的准确性。

In view of the problems of low robustness and complex boundary regularization in the existing window extraction algorithms, a new window boundary extraction method based on grid is proposed. Firstly, the PCA algorithm is used to calculate three feature values in local neighborhood, and the point cloud of building wall is extracted completely according to the relationship between the three feature values;Secondly, the wall point cloud is divided into grid and binarization, and the window boundary is extracted roughly;Then, based on the statistical analysis of the rough extraction window boundary grid, the window boundary is extracted accurately and the grid is established twice;Finally, the sub window area of the same window is combined to extract the window boundary. The accuracy of extracting window boundary by the method is verified by example.