Partial Surface Identification Based on 3-D Multiple Spatial Filtering

This paper proposes a new feature extraction technique for a 3-D surface, based on multiple spatial filtering, and also shows the usefulness of applying the feature to partial shape identification and pose estimation. This method incorporates the trajectory of a point on a surface caused by multiple filtering as characteristics of the point, and can be used robustly for searching feature points and estimating poses of even curved and complex shapes.