A modeling technique of point cloud based on entity-classification
Jiang Heng-bia · Engineering of Surveying and Mapping · 2014
It's difficult to extract the feature from 3D terrestrial laser scanning point cloud because the data is mass and geographic entity information is implied.Based on the characteristics classifying the point cloud into entity is proposed with seed-grown method first.After classification and layering,RANSAC algorithm and robust modeling method can be used to extract the object feature.As for the complexity of the objective world,the use of some prior knowledge can reduce the manual processing time when classification and modeling are done.