RefineCombined

Parameters for iterative “combined” refinement of detected hypotheses.

Such refinement interleaves the 2D-search (LinMod) with the 3D-optimization (an iterative least-squares optimization with the point-to-plane distance metric). One should apply the iterative combined refinement to a search for flat objects, where almost all vectors normal to a surface have similar directions. For such objects, the original 2D-search results in the detections of hypotheses with significant error margins. To refine such hypotheses, we repeat the search by using only the modalities for 2D and 3D edges. One can also iteratively decrease the spread size for these modalities.

Format

Object

Please refer to the sub node documentation.