point cloud to range image to octree

asked 2014-08-02 12:10:46 -0600

sabastian gravatar image

Im trying write serve some processed depth data to a javascript client. My depth data is taken from a kinect to construct a point cloud via PCL. From here I'd like to extract from meaningful keypoints, vertices or borders, compress the pointcloud, add the keypoints to the compressed data in a meaningful way and serve this to the client.

I dont care much about the resolution of the data, just enough for the client to glean a basic geometric representation of the space. Im looking at octree compression of the point cloud via PCL and keypoint extraction from the point cloud via methods which operate on PCL's range image object created from a point cloud, such as NARF keypoint extraction.

However, Im at a loss as to how to map something like the keypoint extraction data from the range image to an octree. Is there a common approach to 'marking' an octree with data like this or moving from a range image to an octree?

edit retag flag offensive close merge delete