Automatically Discovering Urban Features for 3D City Modeling

PhD Thesis Proposal Defence


Title: "Automatically Discovering Urban Features for 3D City Modeling"

by

Mr. Peng ZHAO


ABSTRACT:

Growing 3D map services drives tremendous demand for photo-realistic modeling 
of cities from images captured at ground level. This modeling of cities reduces 
de facto to that of building facades. The accurate extraction and partition of 
individual facades and the semantic analysis of each individual facade are two 
main challenges.

The key to solve these problems is using urban features: rectilinearity and 
symmetry. First, a joint 2D-3D methods assuming rectilinear boundary of facade 
parses the environment into buildings, the ground, and the sky; for the first 
time, buildings are further partitioned into individual facades using a dynamic 
programming optimization. The next step detects and segments structural 
elements in individual facade by exploiting the information redundancy of 
repetition. We propose a dual image- and transform-space optimization method, 
capable of simultaneously discovering multiple interfering repetitions, and a 
per-pixel symmetry segmentation method that can extract repetitive elements 
from background walls. Extensive comparison and evaluation demonstrates the 
significant improvement of our methods over the state-of-the-art. Finally, 
extracted architecture elements are re-assembled into a newly invented grammar 
rules, and several promising applications including façade modeling, texture 
synthesis, façade repair and completion based on the analysis results are 
demonstrated.


Date:                   Wednesday, 6 June 2012

Time:                   1:30pm - 3:30pm

Venue:                  Room 3315
                         lifts 17/18

Committee Members:      Prof. Long Quan (Supervisor)
                         Dr. Pedro Sander (Chairperson)
 			Dr. Chiew-Lan Tai
 			Prof. Chi-Keung Tang


**** ALL are Welcome ****