- Title Information
- Title
- Smooth Signed Distance (SSD) Surface Reconstruction Framework
- Name:
Personal
- Name Part
- Calakli, Fatih
- Role
- Role Term:
Text
- creator
- Name:
Personal
- Name Part
- Taubin, Gabriel
- Role
- Role Term:
Text
- Advisor
- Name:
Personal
- Name Part
- Ritchie, Daniel
- Role
- Role Term:
Text
- Reader
- Name:
Personal
- Name Part
- Harrison, Matthew
- Role
- Role Term:
Text
- Reader
- Name:
Corporate
- Name Part
- Brown University. Engineering: Electrical Sciences and Computer Engineering
- Role
- Role Term:
Text
- sponsor
- Origin Information
- Copyright Date
- 2021
- Physical Description
- Extent
- 15, 126 p.
- digitalOrigin
- born digital
- Note:
thesis
- Thesis (Ph. D.)--Brown University, 2021
- Genre (aat)
- theses
- Abstract
- This thesis presents a global surface reconstruction framework that uses outputs of popular 3D shape acquisition methodologies (ranging from time-of-flight cameras, 3D laser scanners, structured light systems to multi-view stereo algorithms) as input, and generates detailed polygon mesh surfaces. As part of the framework, novel variational formulations are introduced targeted to reconstruct watertight surfaces from various 3D data representations such as oriented points, colored oriented points, calibrated depth maps, cross-sectional curves (parallel or non-parallel), and volumetric occupancy grids. The variational formulations proposed in different chapters are based on minimizing energy functionals composed of simple geometric constraints to estimate a surface implicitly defined as the zero level set of the Euclidean signed distance function to the surface, \ie the implicit function is forced to be a smooth approximation of the signed distance function to the surface. In contrast, the Poisson surface reconstruction approach forces the implicit function to approximate the indicator function of the volume bounded by the implicit surface. Since an indicator function is discontinuous, its gradient does not exist exactly where it needs to be compared with the normal vector data. The smooth signed distance has approximate unit slope in the neighborhood of the data points. As a result, the normal vector data can be incorporated directly into the energy function without implicit function smoothing. In detail, each energy functional consists of two data terms, and one regularization term to constrain the space of signed distance functions. The first data term forces the implicit function to be close to zero at the data points. The second data term forces the gradient of the function to be close to the corresponding normal vectors at the data points. The regularization term forces the gradient of the function to be constant away from the data points. The formulations allow for a number of different discretizations, reduce to scalable least squares problems for all linearly parameterized families of functions, and do not require boundary conditions. The resulting algorithms are significantly simpler and easier to implement, and efficient space-adaptive (octree-based) implementations are shown to produce high-quality adaptive manifold polygon meshes comparable with state-of-the-art algorithms.
- Subject
- Topic
- 3D vision
- Subject
- Topic
- implicit function fitting
- Subject
- Topic
- variational methods
- Subject
- Topic
- shape acquisition
- Subject
- Topic
- polygon mesh
- Subject
- Topic
- adaptive
- Subject
- Topic
- regularized regression
- Subject
- Topic
- surface reconstruction
- Subject
- Topic
- octree
- Subject
- Topic
- watertight surface
- Subject
- Topic
- sparse system of equations
- Language
- Language Term (ISO639-2B)
- English
- Record Information
- Record Content Source (marcorg)
- RPB
- Record Creation Date
(encoding="iso8601")
- 20210607
- Type of Resource (primo)
- dissertations