A REAL TIME IMPLEMENTATION OF 3D SYMMETRIC OBJECT RECONSTRUCTION
MetadataShow full item record
KinectFusion is a surface reconstruction method to allow a user to rapidly generate a 3D model of an indoor scene by moving a standard Kinect camera. With a GPU-based pipeline, the system takes in live depth data from a moving Kinect camera and can create a high-quality 3D model in real time. In this thesis, in order to increase the utility of 3D reconstruction for the symmetric objects, we augment the reconstruction algorithm by adding an occupancy mapping step on top of the surface reconstruction. This is achieved by combining the GPU-based KinectFusion with the CPU-based occupancy mapping algorithm Octomap. The resulting system runs in real time and is used as an input to the symmetry detection algorithm. Tested by the Kinect video stream, our results demonstrate an accurate colored 3D surface reconstruction, which is useful for symmetry detection.