I am calculating the convex hull points array from a 3D model projecting it into a plane. From this convex hull I am calculating the minimum area rectangle enclosing the convex hill. This minimum area rectangle is used to get 2D bounding box for the convex hull. Then I am translating the four points of minimum area rectangle to obtain coordinates of 3D bounding box. But the problem is the direction from which I am obtaining the convex hull is not proper because of which resulting 3D bounding box is not optimum.
Is there any algorithm to obtain the direction for calculating proper convex hull?
There is some missing information in your question. For instance, is your final goal to find the minimum area bounding box of the 3D model?
If that was the case, I would suggest a different algorithm/strategy. There ARE 3D convex hull algorithms implemented, in the CGAL library, for instance. When you go from your 3D model and project it to a plane, you lose a dimension and some crucial information, and from the information presented here I am not sure for what purpose you would pursue such a strategy.
To sum up, the last line doesn't make much sense -- there's no "direction for computing proper convex hull." Convex hulls are computed in 2D, 3D, 4D, etc., no with respect to a particular direction. Good luck!
I guess that you can use Matlab. Have a look at this document:
http://www.appstate.edu/~marshallst/GLY3455/lectures/11_Advanced_Plotting.pdf
I suppose your real task is to find an oriented bounding box with minimal volume (MVB). The implementation of such is not trivial, take a look at David Eberly's MVB implementation (see link below) that is based on O'Rourkes method that utilizes the convex hull for finding the MVB. For finding the 3D convex hull you can use qhull.
http://www.qhull.org/
https://www.geometrictools.com/Samples/Geometrics.html#MinimumVolumeBox3D
AmmyTabb. Thank you for the answer. I tried using Joseph O'roukes 3D bounding box algorithm with time complexity O(n cube). It is time consuming process for large size models which yields greater number of points data. And computing convex polyhedron for a 3D model is also a difficult task which requires accuracy.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Computer Graphics and Animation
- 3D