The Method Of Airborne Laser Scanning Data Classification

The airborne laser radar technology is an emerging technology. Currently, it is used in terrain mapping, urban3D modeling application and so on. This thesis discusses the filtering classification and building point cloud extract algorithm. This thesis describes several laser radar technology filtering algorithm and buildings extraction algorithm. The filtering effect of traditional mathematical morphology, improved mathematical morphology and desired filter are verified by experiments.Traditional slope filtering algorithm uses a priori knowledge to determine the filter threshold, can not get the real terrain slope. In the mountains, the difference between the slope of the terrain slope with little feature is not big. It is difficult to distinguish ground point and the feature point by slope. This will lead to a greater probability of misclassification.To solve the above problem, this thesis presents an improved adaptive gradient algorithm based on traditional filtering algorithm, an increase of two-way slope threshold, elevation threshold and the second filter. Initial DEM in first filtering is produced with smaller slope threshold, and then calculate the maximum slope. The maximum gradient value is used as the slope threshold of the second filter to solve the adaptive problem. One-way threshold in mutations terrain areas is easy to over filtered, an increase of two-way threshold, the general applicability of changes in the slope of the larger region. By increasing elevation threshold we can strengthen the protection of detail.This thesis made some improvements in the research building regional growth algorithm on the basis of the existing regional growth algorithm. After elevation threshold classification, gradient classification and the radius of neighborhood density classification, the seed points of the building will be extracted, the algorithm removes vegetation point and low ground appurtenances interference, solves the over buildings and vegetation separation, the use of the radius of the neighborhood density method can be faster to achieve the classification of the building of the separation.Using the C#.NET and ObjectARX technology, this article develops a secondary development of software in AutoCAD called LIDAR-SCANE. It can display with classified point cloud, build triangulated irregular network and classify point cloud and other functions.