ShapeReconstructionPaLS

This is a flexible framework for multimodal 3D shape reconstruction using parametric level set methods.

The package is based on the following paper (please cite if you are using the package):

Moshe Eliasof, Andrei Sharf, and Eran Treister, Multi-modal 3D Shape Reconstruction Under Calibration Uncertainty using Parametric Level Set Methods, 2019. (https://arxiv.org/abs/1904.10379)

This package relies on two other packages:

1. jInv.jl - A juliA package to solve parameter estimation problems efficiently and in parallel For more details see (http://arxiv.org/abs/1606.07399).
2. ParamLevelSet.jl - A julia package to implement parametric level set methods.

Overview

ShapeReconstructionPaLS consists of three submodules:

1. DipTransform - For shape reconstruction from dipping objects in water in different angles, exploiting Archimedes Law.
2. ShapeFromSilhouette - A no-fill implementation for surface reconstruction from silhouette with support to camera calibration fixup.
3. PointCloud - For surface reconstruction from point clouds with support to registration fixup.
4. Direct - A module to find a parametric level set representation for a given object. For demonstrating the expressiveness of the parametric level set representation.
5. Utils - utility functions

Examples

Some examples can be found in the driver folder. See the results section in https://arxiv.org/abs/1904.10379.