The Division Of Point Cloud Based On Oblique Photogrammetry

Point cloud data have a great prospect on extracting buildings, measuring the height of the trees, acquiring the high-precision digital elevation model, feature monomeric. At present,the method that acquire point cloud data is divided into two categories, one is point cloud data acquired by LiDAR technology, it have applications narrow,long period,low efficiency, and it does not apply to terrain information collection of the large areas. The other one is matching dense point cloud data by oblique photogrammetry technology,apply to terrain information collection of the large areas. The UAV oblique photogrammetry technology is rapidly developed in recent years, it have low cost, simple operation, meeting the demand for mapping large-scale features.At present, there are many algorithm in point cloud data classification, but due to the complexity of the terrain, the result of these algorithm is not good. And the existing point cloud data classification algorithm is mainly used in laser radar scanning point cloud data.There have not an experiment in matching dense point cloud data. For these problems, this article improve the existing point cloud classification algorithm, and this algorithm is appled to matching dense point cloud data. This article mainly completed the following contents:1.The detailed description the composition of the UAV oblique photogrammetry system and mainstream the multi-angle camera in Chinese and foreign, introduces the characteristics of the oblique image. On this basis, we summarized the reason that the oblique photogrammetry does not apply traditional oblique photogrammetry measurement system.2.Briefly describ the process of the point cloud data and the key technology of the process. Based on the principle of image Matching, we summarize the characteristics of the dense point cloud data that obtained by the technology of matching. It describes the principle of the point cloud data classification and these international classical algorithms.This artical summarize the advantages and disadvantages of these algorithms.3.On the basis of the classical filter, combine the image segmentation algorithm for sorting and elevation TIN encryption methods for intensive match point cloud data division.After loading the intensive matching dense point cloud data, the error points of the point cloud data are removed and construct the virtual grid, generating the depth image, and then divided the point cloud data into the ground point and non-ground points by the image segmentation algorithm. At last, the ground points is used to TIN encryption of Elevation sorting, the non-ground points used triangulation filtering with great threshold. Finally, the article reach the ground point and non-ground points precision classification.