medial axis extraction of 3d objects through radial basis functions ming ouhyoung, professor...

13
Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan University

Upload: sydney-hudson

Post on 17-Jan-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Medial Axis Extraction of 3D Objects Through

Radial Basis Functions

Ming Ouhyoung, ProfessorCommunication and Multimedia LabDept. of CSIE, National Taiwan University

Page 2: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Introduction We use radial basis function to extract the

skeleton (medial axis) of an arbitrary polygonal model

Medial axis transformation (MAT) is the locus of centers of all spheres which have maximal radius in the interior of an object.

Page 3: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Ref: Skeleton Extraction of 3D Objects with Radial Basis Function, F-C Wu, W-C Ma, M. Ouhyoung, paper in preparation. Submitted to ACM Multimedia 2002.

Page 4: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Radial Basis Function The shape of a polygon model can

be defined by a radial basis function A simple radial basis function is 1/r3

1

( ) ( ) ( )n

j jj

f x d x c P x

Page 5: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Radial Basis Function RBF is differentiable, which is a good

property for gradient based search algorithms

RBF constructs a kind of level surface for shrinking purpose

Page 6: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Surface Shrinking Process

Local Maximum

Surface Vertex

Page 7: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Surface Shrinking Process For each vertex on the model

surface, let it move while following the gradient until a local maximum is found

Group and connect the local maximum

Page 8: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Medial Axis Construction Algorithm Connnect two local maxima with a line For each sample vertex on the above lin

e, move to the local ridge (maximum) to create a medial axis

Page 9: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Medial Axis Construction Algorithm: Hill Climbing, Ridge Finding

Ridge finding

Axis

Page 10: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Results: the skeletal structure

Two smoothed mesh model of animals: see DEMO (RBFdemo)

Page 11: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Results 1

Page 12: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

Results 2

Page 13: Medial Axis Extraction of 3D Objects Through Radial Basis Functions Ming Ouhyoung, Professor Communication and Multimedia Lab Dept. of CSIE, National Taiwan

The End