|
As the population density of urban areas, areas with a concentration of production and living, and the building model the effective feedback cities spatial information, therefore, select an efficient model of the building extraction method has certain practical significance. Airborne LIDAR technology is a novel measurement techniques, this technology can be fast, accurate extraction of surface LIDAR point cloud data, which can be extracted building models of the urban areas. LIDAR technology to extract building models compared with the traditional aerial image extraction method has many advantages: First, LIDAR is an active sensor, without the aid of other light sources, the less the impact of the external environment, and therefore able to work full time, optical measurement method due to the impact of the external environment can not work full time; Secondly, the LIDAR system can obtain dense point cloud data (5-10 points) per square meter, making the accuracy of the data extracted the edge of the high and characteristic line is more stable, and the characteristic line of the buildings in the optical image in the interference of light and shadow will be rendered unstable characteristics; Again, for those with the optical measuring method is difficult to distinguish the area, such as heavily forested areas, dense urban areas, desert areas, using LIDAR technology to better distinguish. Taken together, the airborne LIDAR technology to extract building models is an accurate and efficient way. View of a limited number of fully automated building extraction algorithm, which largely limits the application of airborne LIDAR technology. Certain key technologies for the buildings based LIDAR point cloud data extraction process, a new automatic extraction method, the specific content of the study include: 1. LIDAR technology knowledge, including the LIDAR system works, the applications of LIDAR point cloud data features, as well as several typical LIDAR point cloud data filtering processing algorithms and building extraction method. 2. Based bands by classification. Since the building points belong to the non-ground point, it is during the process of building extraction, the first thing to do is the original point cloud data classification, dividing it into the ground points and non-ground point two categories. Specifically, the block data strip divided in the X direction and Y direction, respectively, and each point in the point cloud to be classified in these two directions, after classification can be divided into the ground point and non- candidate points on the ground, after screening the candidate of the non-ground points, a point twice the classification results are non-ground candidate point, this point is that the non-ground points to the contrary the ground point, two the Classification different point usually in In this paper, on the steep slopes classification method based zonation based strip divided reduce the dimension of the 3D LIDAR point cloud data, accelerate the classification speed. The experimental results show that, compared to the existing progressive triangulated irregular network encryption method and slope-based filtering method, lower total error rate of the data obtained in the paper, the results are more accurate. The 3. Segmentation and edge extraction based on Delaunay triangulation buildings. LIDAR point cloud data is classified as ground points and non-ground points, a building point edge point should be identified. Extracting points belonging to a building known as the split, most of the existing buildings segmentation method need to select seed point, not yet particularly desirable seed point selection method. In addition, most of the existing edge detection method based on the method of edge detection of the image, and extracted to the edge of the building and the actual edge of these methods exist differences in a certain position and therefore used for the image's edge extraction operator usually only The GRID format data. In view of the deficiencies in the existing buildings segmentation and edge extraction method for irregular spatial distribution of discrete LIDAR point cloud data segmentation and edge extraction, the need to seek other solutions. In this paper, the building of a suitable discrete LIDAR point cloud data segmentation and edge extraction method. Delaunay triangulation Delaunay triangulation of LIDAR point cloud data, in order to establish the relationship between points adjacent split piece then judge each triangle in the triangulation network belongs Finally, the area of ??control to eliminate debris, thereby removing the noise of the small area of ??the building height Grove. Select the Changchun City Culture Square area of ??four square kilometers of LIDAR point cloud data as the experimental data to validate the proposed method. Experimental results show that the use of this method can be detected in the vast majority of the building point in the point cloud data, and the edge line of each building can be extracted, and the extracted edge line is relatively smooth. Visible, the proposed method can meet the needs of general building extraction.
|