Computational Shape Understanding for 3D Reconstruction and Modeling
EPFL PhD Thesis No. 6259
The physical and the digital world are becoming tightly connected as we see an increase in the variety of 2D and 3D acquisition devices, e.g., smartphones, digital camera, scanners, commercial depth sensors. The recent advances in the acquisition technologies facilitate the data capture process and make it accessible for casual users. This tremendous increase in the digital content comes with many application opportunities including medical applications, industrial simulations, documentation of cultural artifacts, visual effects etc.
The success of these digital applications depends on two fundamental tasks. On the one hand, our goal is to obtain an accurate and high-quality digital representation of the physical world. On the other hand, providing a high-level understanding of the digital content similar to a human being is crucial. Both of these tasks are extremely challenging due to the large amount of available digital content and the varying data quality of this content including noisy and partial data measurements. Nonetheless, there exists a tight coupling between these two tasks: accurate low-level data measurement makes it easier to provide a high-level understanding of the digital content, where as use of suitable semantic priors provides opportunities to increase the accuracy of the digital data.
In this dissertation, we investigate the benefits of tackling the low-level data measurement and high-level shape understanding tasks in a coupled manner for 3D reconstruction and mod- eling purposes. We specifically focus on image-based reconstruction of urban areas where we exploit the abundance of symmetry as the principal shape analysis tool. Use of symmetry and repetitions are reinforced in architecture due to economic, functional, and aesthetic considera- tions. We utilize these priors to simultaneously provide non-local coupling between geometric computations and extract semantic information in urban data sets.
Concurrent to the advances in 3D geometry acquisition and analysis, we are experiencing a revolution in digital manufacturing. With the advent of accessible 3D fabrication methods such as 3D printing and laser cutting, we see a cyclic pipeline linking the physical and the digital worlds. While we strive to create accurate digital replicas of real-world objects on one hand, there is a growing user-base in demand of manufacturing the existing content on the other hand. Thus, in the last part of this dissertation, we extend our shape understanding tools to the prob- lem of designing and fabricating functional models. Each manufacturing device comes with technology-specific limitations and thus imposes various constraints on the digital models that can be fabricated. We demonstrate that, a good level of shape understanding is necessary to optimize the digital content for fabrication.