IRIS Research Staff

Yiyong Sun
Ph.D. Student

Office: 410 Science & Engineering
The University of Tennessee
Knoxville, TN 37996-2100
Telephone: (865) 974-9213
Fax: (865) 974-5459
E-mail: yiyong@iristown.engr.utk.edu
Personal Web Page: /~yiyong/


Dissertation Work: Surface Modeling and Processing From Multiple Range Images

The research addresses several problems in the process of reconstructing surfaces from multiple range images. The first is smoothing the raw surfaces scanned by the range scanners. The second is finding registration between pairs of range images. In the third step, surface mesh is reconstructed by integrating range images from different views. Segmentation is the last step that partitions the 3D mesh into meaningful parts.

The techniques developed in this research include:

The proposed algorithms can be applied in various applications including reverse engineering, scene building and understanding, virtual reality, and object recognition.


Surface Smoothing by Area Decreasing Flow - As an alternative to mean curvature flow, local area minimization avoids curvature estimation and computes an optimal magnitude of surface deformation.
Surface Registration by a Point's Fingerprint Matching - A point's fingerprint is defined at every 3D point. Surfaces are registered by finding the corresponding points based on fingerprint comparison.
Surface Reconstruction Using an Implicit Surface-Based Method - Signed distance fields are constructed from each triangle mesh generated by range image. After the integration of the distance field, the surface, along with the texture, can be extracted.
Surface Segmentation by 3D Watershed - Edge strength is approximated at each vertex. The piecewise continuous edge strength map is used to partition the triangle mesh using a watershed algorithm.


Last updated:
Webmaster