Skip to content
pcl point cloud registration
C++ Python C
Find file
New pull request
Fetching latest commit...
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.



working with openPCL for point cloud segmentation, the workflow is as follows

1) obtain point clouds from camera / projector system 
	 merge and register these clouds (not done yet)
	 pcd_write will create a pcd file from a packed binary depth map, this file is
	 defined as: <I:npts> <I:height> <I:width> npoints*{<f:px> <f:py> <f:pz>}
2) segment full point cloud into separate clouds each one representing
	 a cube
	 cloud cluster will read a pcd file and do this arguments are: inputfile outputpath

3) for each segmented file attempt to figure out the actual cube it represents 
	 in the world space. There are a variety of putative methods
	 a) ransac as many planar models as possible to the point cloud, removing
	 		inliers each time. Then try and compute the verts / edges of the cube by 
			figuring out all the 2 / 3 plane interesections from the set
	 b) compare cube methods, generate a substitute cube. Given the camera 
	 position in the world we can estimate the pitch of a square pixels on the face
	 (supposing that the cube is head on). Use this to generate another point cloud
	 representing the subsitute cube. Use ICP or cleverer methods to register
	 this cloud with the test cloud, this gives the centroid and the orientation.
Something went wrong with that request. Please try again.