This Project was a part of 3D Vision course at ETH Zurich

• tested under ubuntu 16.04
• requires cmake, boost and OpenCV
• you can use an IDE such as QtCreator to make your life easier

Run the following commands:

cd vision-normal-fusion mkdir build cd build cmake .. make -j4

Example for the cube dataset "datasets/cube"

• "depthMaps" contains the depth maps and normals (raw and visualization)
• use --help to get some information about the arugments

To integrate the depth maps run the following commands (from "datasets/cube"):

ls depthMaps/depth*.dat > depthMaps.txt ../../build/tvHistDepthIntegrator --configFile conf.txt --depthMapListFile depthMaps.txt --outputGridFile intCSDFGrid.dat

To run the TV-Hist algorithm run the following command:

../../build/tvHistFusion --configFile conf.txt --intDepthGridFile intCSDFGrid.dat --vrmlOutputFile tvHistModel.wrl

open the tvHistModel.wrl file using MeshLab to see the result.

To integrate the normals run the following commands:

ls depthMaps/normals*.dat > normals.txt ../../build/tvHistNormalIntegrator --configFile conf.txt --depthMapListFile depthMaps.txt --normalMapListFile normals.txt --outputFolder integratedNormals

To integrate the normals run the following commands:

ls integratedNormals/normals*.dat > normalsFiles.txt ../../build/tvHistNormalClustering --configFile conf.txt --normalsListFile normalsFiles.txt --outputFile clusteredNormals.dat

To run a TV-Hist algorithm that takes the normals into account run the command:

../../build/tvHistNormalFusion --configFile conf.txt --intDepthGridFile intCSDFGrid.dat --clusteredNormalsFile clusteredNormals.dat --vrmlOutputFile tvHistNormalClusteringModel.wrl