I am working on 3D reconstruction algorithms for a project. I frequently come over the topics "ray-casting" and "marching-cubes" in the 3D surface reconstruction literature. I read that ray-casting on a "signed distance function (SDF)" representation of 3D point cloud creates a surface model. But I think I have not understood this well. Does ray-casting produce a 3D surface model? I know that ray-casting helps us find the zero crossing in a SDF volume and thus we will get the surface points. But does ray-casting create a mesh like 3D model like the marching cubes algorithm? Please suggest me a good document where I can lean about ray-casting used in surface reconstruction.
Asked
Active
Viewed 828 times
2 Answers
2
No, ray-casting doesn't create a model. It can just be used to temporarily visualize (render) your model, but it won't give you a polygon mesh. This might be interesting for real-time feedback like required in Kinect Fusion and similar applications. If you want a polygon mesh, you'll have to use marching cubes on the SDF.
Jochen S
- 36
- 1
0
You could check with a novel method, "Cone Carving for Surface Reconstruction", which uses the concept of cone-visibility of point that's akin to raycasting.
Kabu
- 519
- 6
- 16