-
[show abstract]
[hide abstract]
ABSTRACT: This article proposes a method for estimating the shape of masonry elements present in the facade of a Gothic building from a single image. Our approach takes as input a rectified image of a Gothic building facade and user-specified side information and provides a 3D model estimate of structural elements, e.g., doorways, windows, arches and cornices, within the facade as output. Facade estimation proceeds in two steps: (1) estimation of arches and rectangular openings and (2) estimation of the masonry, i.e., mortar and bricks, surrounding these structures. Arches and rectangular facade elements are detected and extracted using a 2-pass algorithm. Pass 1 detects and estimates individual facade elements using active contours with shape-preserving constraints. Pass 2 groups elements based on their shape similarity, proximity, and horizontal and vertical positions. Pass 1 and 2 are iterated multiple times to extract hierarchical arrangements, i.e., arches within arches that are typical to Gothic architecture. Those pixels not included as part of the architectural elements are considered masonry and are segmented into two classes: (a) mortar and (b) bricks. While current techniques use 3D scans or over-simplify facades using generic 3D models and texture-on-plane methods, the proposed work establishes promising initial steps towards estimating a brick-and-mortar model from imagery alone, i.e., a model of the actual facade components. Such models can expedite preservation efforts by providing detailed records of the geometry of these structures which may collapse or require repair and provides quantitative measurements of building components for use in research on the methods and tools used to construct these buildings.
Computer Vision and Pattern Recognition Workshops (CVPRW), 2010 IEEE Computer Society Conference on; 07/2010
-
E. Gay,
D. Cooper,
B. Kimia,
G. Taubin,
D. Cabrini,
S. Karumuri,
W. Doutre,
S. Liu,
K. Galor,
D. Sanders, A. Willis
[show abstract]
[hide abstract]
ABSTRACT: REVEAL, (Reconstruction and Exploratory Visualization: Engineering meets Art / ArchaeoLogy) is a four year NSF-funded project promoting paradigm shifts in archaeology, currently at the 1.5 year point This is a project to create an environment for acquiring and presenting archaeological data in a way that streamlines the excavation process and supports and enhances the expert's understanding of the data. REVEAL leverages three aspects of technology: using vision algorithms to speed up or replace measurement and documentation tasks, using computer automation to speed up data entry tasks, using integrated 2D and 3D media to enhance data comprehension. This paper is an update on what the project has accomplished, what has been learned, and what is planned for the rest of the project.
Computer Vision and Pattern Recognition Workshops (CVPRW), 2010 IEEE Computer Society Conference on; 07/2010
-
[show abstract]
[hide abstract]
ABSTRACT: This article describes a novel method for localization of a robot within a 2D scene given a binary map of the scene and a set of range measurements obtained by the robot from some unknown position and orientation. Theoretically, the algorithm is capable of solving all recognized variants of the robot localization problem: tracking, global localization, and kidnapped robot. This is accomplished by treating each set of range measurements as a unique fingerprint, referred to as a range pattern, that is associated with each potential (x, y, ¿) pose of the robot. We provide detailed theoretical analysis and an exact solution for the problem when both the range and angle measurements are constrained to come from a discrete set of possible values. Experimental results are obtained using simulated range data taken from synthetic and real-world maps to provide insight on the robustness of our approach and identify situations where the localization solution obtained is not unique. Our solution to this more-constrained problem has low computational complexity and is exact which makes it appropriate for use in real-time robotic navigation applications. Solutions to this problem are of great importance for successful deployment of autonomous robotic vehicles within a-priori known spaces, e.g., buildings, hospitals, etc.
IEEE SoutheastCon 2010 (SoutheastCon), Proceedings of the; 04/2010
-
[show abstract]
[hide abstract]
ABSTRACT: This paper presents a system for virtual reconstruction of comminuted bone fractures. The system takes as input a collection of bone fragment models represented as surface meshes, typically segmented from CT data. Users interact with fragment models in a virtual environment to reconstruct the fracture. In contrast to other approaches that are either completely automatic or completely interactive, the system attempts to strike a balance between interaction and automation. There are two key fracture reconstruction interactions: (1) specifying matching surface regions between fragment pairs and (2) initiating pairwise and global fragment alignment optimizations. Each match includes two fragment surface patches hypothesized to correspond in the reconstruction. Each alignment optimization initialized by the user triggers a 3D surface registration which takes as input: (1) the specified matches and (2) the current position of the fragments. The proposed system leverages domain knowledge via user interaction, and incorporates recent advancements in surface registration, to generate fragment reconstructions that are more accurate than manual methods and more reliable than completely automatic methods.
Computer Vision Workshops (ICCV Workshops), 2009 IEEE 12th International Conference on; 11/2009
