Registration of images to unorganized 3D point clouds using contour cues
Resource Type | Date |
---|---|
Software | 2016-09-08 |
Description
The following github repository contains the code developed in this project: https://github.com/imatge-upc/multimodal-registration
About
This project contains the tools necessary for testing the contour-based multimodal registration algorithm.
Set-up
The code provided contains a Dockefile containing all the necessay dependencies to compile the project. Scripts build.sh
and run.sh
are also provided to build the image and run a container with the specified images. Any environment variable stating with PHD_*
will be autmatically added to the container. Please configure your datasets and results folder in the run.sh
script.
Provided tests
This software is shipped jointly with a demo test of the algorithm. For running this test you will need the Stanford bunny data, both an static image of the bunny and a 3-D reconstuction. This test can be found into the autorun
folder. To run this test, run python autorun/runners.py fullpath
on your machine. Additionally, a test on the color gradient algorithm is also provided. To run this test, run python autorun/runners.py gradientTestSynth
.
You can find the steps done on each test in the scripts autorun/fullpath.py
and autorun/gradientTestSynth
.
People involved
Alba Pujol | PhD Candidate |
Javier Ruiz Hidalgo | Associate Professor |
Josep R. Casas | Associate Professor |
Related Publications
“Correspondence matching in unorganized 3D point clouds using Convolutional Neural Networks”, Image and Vision Computing, vol. 83-84, 2019. (3.95 MB) | ,