Skip to content

p1lare/housescan

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

229 Commits
 
 
 
 

Repository files navigation

HouseScan

A distinguished Computing Master's project at Imperial College London, 2014. Supervised by Prof. Andrew Davison. Written in Haskell. Read the thesis or watch a video.

Building-scale interior 3D reconstruction with KinectFusion

You can use HouseScan to create a 3D model of your house, flat, or anything that has rooms aligned in a grid.

Usage

  • Scan each room with my version of KinFu from my PCL fork
  • Perform plane detection with the the plane detection tool I added to that fork
  • Import the rooms into the HouseScan, arrange them
  • Export them in full resolution as .ply files
  • Load them all into Meshlab to see the result

You can find detailed instructions in the appendix of my thesis (linked below), section A HouseScan user manual.

Prerequisites

For a full scan you need:

  • An RGB-D camera: Asus Xtion Pro, Microsoft Kinect, or similar camera working with OpenNI
  • A computer with a fast Nvidia graphics card supporting CUDA (preferably a laptop for carrying around) - I used a GTX 780M
  • Linux installed (preferably Ubuntu 14.04 or equivalent)

License

HouseScan is MIT licensed. PCL/KinFu are BSD licensed. You can use them free of charge, but please tell me when you do something cool with it!

About

Building-scale interior 3D reconstruction with KinectFusion

Resources

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages

  • Haskell 100.0%