CPC G06T 17/20 (2013.01) [G06T 7/50 (2017.01); G06T 2207/10028 (2013.01)] | 19 Claims |
1. A method, comprising:
receiving, from a capturing device, one or more frames of depth data;
generating, from the one or more frames of depth data, a 3D point cloud representation of a 3D space;
creating an existing set of voxels which encompass the 3D point cloud;
receiving, from the capturing device, a new frame of depth data;
generating, from the new frame of depth data, a new 3D point cloud representation of the 3D space;
determining a number of new voxels needed to encompass points of the new 3D point cloud;
identifying voxels of the existing set of voxels having depth values to be replaced by corresponding depth values of the new frame of depth data; and
deciding whether to mark the new frame as a key frame based on the number of new voxels needed to encompass the points of the new 3D point cloud and a number of the voxels for which the depth values are replaced.
|