激光点云技术扫描建筑物门窗立面数据提取

Facade data extraction of building doors and windows scanned by laser point cloud technology

三维快速建模在未来城市作战中起着关键作用,建筑模型信息越细节、越完善,就越能够充分发挥出建筑模型的作用。为此,以激光点云技术为扫描手段,提出一种建筑物门窗立面数据提取方法。利用设计的激光点云设备扫描模式与间距测量方法,采集建筑物点云数据,将其投影至水平网格中,归一化处理各区域灰度值,基于数据点与其邻域点间的欧几里得距离,结合噪声点判定阈值,滤除无效离散点,依据灰度级发生概率与平均灰度值,分割出建筑立面点云,根据门窗特征与立面深度差异,旋转、投影建筑点云数据后得到门窗立面点云数据集合,针对面片类别提取出对应的门窗点云数据。实验阶段中,选用RIEGL VZ-1000三维激光扫描设备采集点云数据,由提取的视觉效果可知,所提方法能够完整输出门窗信息,经对比窗体宽高与数量,验证该方法平均绝对偏差值为1.43 cm,具有较好的提取精准度,最高可达到98%,同时提取误差分布等级分布较低,误差仅为2%。

3 D rapid modeling plays a key role in the future urban combat. The more detailed and perfect the information of building model, the more it can give full play to the role of building model. For this reason, a method of building window facade data extraction is proposed based on laser point cloud technology. Based on Euclidean distance between data points and their neighboring points, combined with the threshold value of noise points, the invalid discrete points are filtered out, and the building facade is segmented according to the probability of gray level occurrence and the average gray value point cloud, according to the differences between the features of doors and windows and the facade depth, the point cloud data set of the facade of doors and windows is obtained after rotating and projecting the building point cloud data, and the corresponding point cloud data of doors and windows is extracted according to patch category. In experimental stage, the RIEGL vz-1000 three-dimensional laser scanning equipment is selected to collect point cloud data. From the visual effect of extraction, the proposed method can output the door and window information completely. By comparing the width, height and number of windows, the average absolute deviation of this method is 1.43 cm, which has a good extraction accuracy, up to 98%, at the same time, the extraction error distribution level is low, and the error is only 2%.