Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Register multiple point clouds into one coordinate system
Fetching latest commit…
Cannot retrieve the latest commit at this time.
|Failed to load latest commit information.|
Author: Nathan Crock email@example.com http://www.mathnathan.com Date: 03/23/11 14:53 ----------- Kinect Camera Registraion ------------ OS SUPPORT: Only tested on Linux so far Ubuntu 12.04, 12.10 FILES: KinReg.cpp - The registraion program CMakeLists.txt - Cmake file with build commands cmake | | find | | Find*.cmake all of the cmake files to find the | dependencies DEPENDENCIES: Freenect OpenCV OpenGL GLUT COMPILATION INSTRUCTIONS: -If you don't have cmake you can get it with aptitude sudo apt-get install cmake -For a breif introduction to CMake follow my tutorial http://mathnathan.com/2010/07/11/getting-started-with-cmake/ -Make a build directory and from it run cmake .. make NOTE: If you have the dependencies installed (to your root directories, or their paths in your ldpaths), then it should "just work" in UBUNTU 12.04 and 12.10. Please let me know otherwise ---------------------------Simple Project Manual-------------------------------------- This program is intended to register multiple point clouds into one coordinate system. The point clouds are obtained from Microsoft's Kinect sensors. NOTE: Only support for 2 cameras for now It begins by finding correspondences between the two cameras. The process is interactive. Once KinReg is running, a window pops up containing both images side by side. All you do is click on pixels in both cameras which correspond to the same point in world coordinates. Currently the process is rough, it can be improved in the future by drawing circles around the points you click on and adding colors to signify if the correspondence is accepted or not. If both depth measurements for your clicks are found it stores the correspondence. If at least one of the corespondences is bad (no depth) it deletes the match. Once you've collected around 10 or so correspondences pushing 'p' applies procrustes analysis on the correspondences and calculates the transformation matrices. Press 't' to see the translation of the centroids to the origin Press 'a' to see the translation and rotation applied to both point clouds ====================================================================================== "Believing in the way, makes the way!"