We present an algorithm for segmenting a mesh into patches whose boundaries are aligned with prominent ridge and valley lines of the shape. Our key insight is that this problem can be formulated as Correlation Clustering (CC), a graph partitioning problem originated from the data mining community. The formulation lends two unique advantages to our method over existing segmentation methods. First, since CC is non-parametric, our method requires few parameters to tune. Second, as CC is governed by edge weights in the graph, our method offers users direct and local control over the segmentation result. Our technical contributions include the construction of the weighted graph on which CC is defined, a speed-up strategy for computing CC on this graph, and an interactive tool for editing the segmentation. Our experiments showed that our method produces qualitatively better segmentations than existing methods on a wide range of inputs.
@article {FeatureSeg17,
title = {Feature-Aligned Segmentation using Correlation Clustering},
author = {Yixin Zhuang, Hang Dou, Nathan Carr, and Tao Ju}
journal = {Computational Visual Media, (Computational Visual Media Conference2017)},
year = {2017},
volume = {3},
number = {2},
pages = {147-160}
}
We thank Dongming Yan for providing the code of [47] for comparison. The models in this paper are obtained from AIM@SHAPE and Princeton Segmentation Benchmark. The work is supported in part by a gift from Adobe System, Inc..