A Novel Technique of Feature Extraction for 3-D Face Recognition
Pankaj Dadhich, Sachin Borse · International journal of advance research and innovative ideas in education · 2018
Various techniques are available for extracting curves form 3D meshes. Instead of extracting curves form 3D meshes directly, in this paper we apply principal mean curvature approach and then compute Discrete Wavelet Transform (DWT) to extract facial profile. This approach is much more efficient and robust compared to extracting curves form the mesh data. Once the Extraction of geometric Curvature information from 3D facial surface is done, then next step is to perform 3D face recognition using LDA, a non-fiducial method as these are often unreliable. The feature level fusion prior to LDA improves the recognition rate considerably. Basel 3-D face scans database is used for demonstrating this approach.