3D reconstruction from silhouettes and photoconsistency based on depth map fusion

Cots Pujol, Helena

3D reconstruction from silhouettes and photoconsistency based on depth map fusion

License

Citation

Cots Pujol, H. 3D reconstruction from silhouettes and photoconsistency based on depth map fusion. 2019. handle: http://hdl.handle.net/10230/42384

This citation was generated automatically.

Abstract

In this thesis, we present a multi-view 3D reconstruction system from silhouettes and photoconsistency information based on depth map fusion. The method receives as input a set of images from different points of view with their corresponding calibration parameters. A confidence volume is extracted from the silhouette information. From this initial volume, photometric consistency is computed for every voxel inside the volume with respect to every view in the dataset using DAISY descriptors. Then, a set of depth maps is built for every view and they are merged via Truncated Signed Distance Function. Finally, the Marching Cubes algorithm is used to extract a polygonal mesh of the 3D implicit surface.