Computer Science and Engineering PhD Thesis Defense by Emel Küpçü



KOÇ UNIVERSITY

GRADUATE SCHOOL OF SCIENCES & ENGINEERING

COMPUTER SCIENCE AND ENGINEERING

PhD THESIS DEFENSE BY EMEL KÜPÇÜ

 

Title: Reliable 3D Shape Matching on Challenging Point Clouds Captured by Commodity Depth Cameras

 

Speaker: Emel Küpçü

 

Time: September 13, 2018, 15:00

 

Place: ENG208

Koç University

Rumeli Feneri Yolu

Sariyer, Istanbul

 

Thesis Committee Members:

Prof. Yücel YEMEZ (Advisor, Koç University)

Prof. A. Murat TEKALP (Koç University)

Assoc. Prof. Engin ERZİN  (Koç University)

Assoc. Prof. Yusuf SAHİLLİOĞLU (Middle East Technical University)

Asst. Prof. Helin DUTAĞACI (Eskişehir Osmangazi University)

 

Abstract:

We propose iterative isometric point correspondence methods that rely on diffusion distance to handle challenges posed by commodity depth cameras, which usually provide incomplete and noisy surface data exhibiting holes and gaps. We formulate the correspondence problem as finding an optimal partial mapping between two given point sets, that minimizes deviation from isometry. We start with an initial rough correspondence between keypoints, obtained via any point matching technique. This initial correspondence is then pruned and updated by iterating a perfect matching algorithm until convergence to find as many reliable correspondences as possible. The resulting set of sparse but reliable correspondences then serves as a base matching from which a dense correspondence set is estimated. For shapes with intrinsic symmetries such as human models, we additionally provide a global intrinsic symmetry technique to increase the resiliency of our correspondence methods against symmetric flips. The experiments show that our methods provide state of the art performance over depth frames exhibiting occlusions, large deformations and topological noise.