University of Waterloo

Fast and Accurate Surface Normal Estimation from Depth Maps

Saed Moradi, PDF

June 17, 2022, 11:30am EC4-2101A

Surface normal vector estimation is a common process in different 3D vision and 3D processing tasks such as 3D surface reconstruction, scene segmentation, object recognition, and others. A complete 3D processing pipeline may very well fail due to the lack of an effective surface normal estimation process, or because a given method fails to be fast enough to be used in practical scenarios.  In this aspect of my research, I will present some recent work that accurately estimates surface normals from depth maps.