5.4. Tensor Voting in 3D
We proceed to the generalization of the framework in 3D. The remainder of this section shows that no significant modifications need to be made, apart from taking into account that more types of perceptual structure exist in 3D than in 2D. In fact, the 2D framework is a subset of the 3D framework, which in turn is a subset of the general ND framework. The second-order tensors and the polarity vectors become 3D, while the voting fields are derived from the same fundamental 2D second-order stick voting field.
In 3D, the types of perceptual structure that have to be represented are regions (which are now volumes), surfaces, curves, and junctions. Their terminations are the bounding surfaces of volumes, the bounding curves ...
Get Emerging Topics in Computer Vision now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
