Centerline and tree branch skeleton determination for virtual objects

1. A computer-based method for determining a centerline through a region of interest in a 3D image dataset comprising:

identifying the boundaries of the region of interest in the 3D image dataset;

identifying the endpoints of the region of interest;

determining a gradient field for each point in the region of interest;

identifying regions of points in the region of interest where the gradient field exhibits a non-uniform gradient; and

connecting the regions having a non-uniform gradient to identify a plurality of points proximate to the centerline of the region of interest;

for those plurality of points proximate to the centerline, determining a penalty value which is a function of proximity of the point to a boundary; and

determining a path connecting the endpoints which has the minimum penalized distance wherein the penalized distance reflects the actual accumulated pathlength and the penalty values associated with the points along the path.

2. The method of claim 1 , further comprising smoothing the path connecting the endpoints.

3. The method of claim 1 , wherein at least one of the endpoints is selected based on prior knowledge of the object.

4. The method of claim 1 , further comprising:

identifying the endpoints of branches from the centerline;

for each identified endpoint, determining a branch path connecting the endpoint to another branch which has the minimum penalized distance wherein the penalized distance reflects the actual accumulated pathlength and the penalties associated with the points along the branch path.

5. The method of claim 4 , further comprising computing a distance from boundary field for each of the points in the branch paths.

6. The method of claim 4 , further comprising, for each branch, identifying points which are near the centerline of the branch and performing the steps of determining a penalty value and determining a path only on the points near the centerline.

7. The method of claim 6 , wherein the operation of identifying point near the centerline of the branch further comprises:

determining a gradient field for each point in the region of interest;

identifying regions of points having a non-uniform gradient; and

connecting the regions having a non-uniform gradient.

8. The method of claim 7 , further comprising smoothing the centerline of each branch.

9. A method of generating a centerline through a virtual object comprising:

acquiring a 3D image dataset of the virtual object, the image dataset comprising a plurality of voxels;

identifying the boundaries of a region of interest in the 3D image dataset;

computing a distance from boundary field for each of the voxels in the region of interest;

determining a gradient voxel field for each voxel in the region of interest;

identifying regions of voxels having a non-uniform gradient;

connecting the regions of non-uniform gradient;

from the connected regions, identify a root voxel;

for the connected voxels, calculate a penalized distance from the root voxel; and

determine the minimum penalized distance from the root voxel to the farthest voxel from the root voxel.

10. The method of claim 9 , further comprising smoothing the path defined by the minimum penalized distance.

11. The method of claim 9 , further comprising:

determining the endpoints of branches;

for each branch endpoint, determining a branch path connecting the endpoint to another branch which has the minimum penalized distance wherein the penalized distance reflects the actual accumulated pathlength and the penalties associated with the points along the branch path.