Shape-Based 3D Surface Correspondence
Using Geodesics and Local Geometry
Yongmei Wang, Bradley S. Peterson and Lawrence H. Staib
Computer Vision and Pattern Recognition, June 2000
| 
| 
|
| 
| 
|
| (a) Scale 1
| (b) Scale 2
| (c) Scale 3
|
Figure 1: Sulci (red) and gyri (green) points of the study brain by
thresholding the curvedness (Eq.()).
Top: dorsal view; Bottom: ventral view.
(a): scale 1 (Gauss smoothing sigma = 0.5),
sulci (35%), gyri (15%);
(b): scale 2 (smoothing sigma = 1.5), sulci (25%), gyri (10%);
(c): scale 3 (smoothing sigma = 2.5), sulci (10%), gyri (3%).
(Note: the percentages shown in the brackets are the number of
selected sulci or gyri
points over the total number of surface points.)
Figure 13: Evaluation of different methods for the known warp
(Eq.(10)).
(a): true absolute distance map and direction of differences
(left and right are two different views);
(b): results by our shape-based method (initial points from known warp)
with absolute distance mean square error 3.66;
(c): results by our shape-based method (initial points
from our point matching procedure)
with absolute distance mean square error 8.68;
(d): results by the simple closest point method
with absolute distance mean square error 16.03.
back to Home Page
