Improved Morphometrics Filtering And Region Grow Method Based Buildings’ Lidar Cloud Points Exploration

As a kind of photogrammetric technique to obtain the precise DEM production in real-time, LiDAR has been widely applied in more and more fields. Simultaneously, it is widely reseached in digital city to acquire the3D informationg about urban buildings. The cloud points attained by LiDAR coule clearly describe the3D information about the interested objects, so it is an important method to get the buildings’characteristic information.To obtain the available information about buildings, the filtering method and building detection are the key points. By summarizing the methods for filtering LiDAR points and building detection, this thesis proposes an improved morphometric filtering method and a region growing method to detect buildings, and verify its feasibility. The main research works contain:(1) To introduce the functionalities and properties about TerraSolid software, and the workflow and related principles to achieve filtering and building detection by TerraScan.(2) To improve the morphometric filtering method, as the classical mathematical morphology using the fixed structure to process, it is cannot be perfectly applied for the objects with sizes changed obviously. So this thesis proposes a method with an variational filtering structure to classify, and verifies its efficiency.(3) Considering the different characteristic parameters between the buildings and vegetations, it could use the region growing method to detect buildings. For the growing method based on TIN, the estimation parameters is the elevation changes between different data types from adjacent parts, but it may result in precision declined. So this thesis proposes that do not apply region growing for the whole cloud points, but to filter the points firstly to ground points and non-ground points., and then to detect buildings from non-ground points, and finally verify its availability.