Consistent segment-wise matching with multi-layer graphs (CGVC2018 Poster)
Taiwei Wang, David Solomon George, Yu‐Kun Lai, Xianghua Xie, Gary K.L. Tam · Cronfa (Swansea University) · 2019
Segment-wise matching is an important problem for higher-level understanding of shapes and geometry analysis.Many existing segment-wise matching techniques assume perfect segmentation, and would suffer from imperfect or over-segmentation inputs.To handle this shortcoming, we propose a multi-layer graph (MLG) to represent possible partially merged segments of input shape.We adapt the diffusion pruning technique on the MLGs to find high quality segment-wise matching.Experimental results on man-made shapes demonstrate the effectiveness of our method.