
Mathematics is wonderful! This video reveals the primary ideas of the feature extraction for plane, sphere, cylinder, cone, and torus from point cloud. A curvature analysis around the pick-point provides the initial surface parameter values. Starting from this initial surface, the Orthogonal Distance Fitting runs round a loop of segmentation and model fitting till some break conditions are fulfilled. Single-point clicking is highly relevant to the automatic feature extraction from point cloud. In contrast, the RANSAC for plane extraction selects randomly three points in a point cloud. The probability that the three points selected belong to an object plane is generally low. Much better is to pick a single-point and then to determine the proper surface type for the point cloud patch around the pick-point. Computer will eventually take the job of single-point clicking over. The prerequisites for a possible fully-automatic feature extraction from point cloud are: 1. Feature extraction directly from point cloud without meshing. 2. Single-point probing (picking, clicking). 3. Robust and stable model fitting algorithm. 4. Broad range of feature types that can be identified and extracted. Orthogonal distance fitting together with differential geometry and some other algorithmic techniques will make a fully-automatic feature extraction possible. System: Windows 7, Intel Core i7, 3.40GHz, 16 GB Memory, NVIDIA GeForce GTX 760. 90: number of the extracted features. 13 min.: working time for mouse clicking and keyboard pressing. 3-D object recognition, pipe, reducer, elbow, least squares, Hough transform, RANSAC, surfcurv, curvsurf.
Feature Extraction from 2 Mio. Points - YouTube |
| Likes | Dislikes |
| 756 views views | 377 followers |
| Science & Technology Creative Commons Attribution licence (reuse allowed) | Upload TimePublished on 28 Jan 2014 |
3d printer,3ds,3d ultrasound,3ds emulator,3d printing,3ds games,3ds xl,3d tattoos,3d pen,3d warehouse,3d pictures,3d tuning,3d nails,3d doritos,3ds max,3d shapes,3d printed house,3d scanner,3d printing service,3dmark
Không có nhận xét nào:
Đăng nhận xét