Shape Clustering as a Type of Procrustes Analysis
The ordinary Procrustes sum of squares is one of the most important measures in Procrustes analysis of shape. In this paper, we incorporate a competitive learning scheme into Procrustes analysis. We introduce a measure of distance between the landmarks of shapes for a competitive learning scheme. Thus, we present novel shape clustering as a type of Procrustes analysis. Using datasets of line drawings and outlines, we show that shape clustering performs well for shape classification compared with shape clustering founded on typical vector-based distances.
KeywordsShape clustering Procrustes analysis Competitive learning
We wish to thank Yoshitaka Toda for his support with the experiments. We thank Maxine Garcia, PhD, from Edanz Group (www.edanzediting.com/ac) for editing a draft of this manuscript. This work was supported in part by JSPS KAKENHI Grant Number JP16K00308 and Hiroshima City University TOKUTEI-KENKYUHI Grant Number 1150347.
- 1.Dryden, I.L., Mardia, K.V.: Statistical Shape Analysis with Applications in R. Wiley Series in Probability and Statistics, 2nd edn. Wiley, Chichester (2016)Google Scholar
- 2.Ferrari, V.: ETHZ Shape Classes, September 2009. http://www.vision.ee.ethz.ch/datasets/index.en.html
- 7.Kuncheva, L.I., Hadjitodorov, S.T.: Using diversity in cluster ensembles. In: Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, pp. 1214–1219. IEEE, The Hague, October 2004Google Scholar
- 9.Okamoto, T., Iwata, K.: Data set of skewed line drawings, September 2017. http://www.prl.info.hiroshima-cu.ac.jp/~kiwata/okamotoiwata/
- 10.Okamoto, T., Iwata, K., Suematsu, N.: Extending the full procrustes distance to anisotropic scale in shape analysis. In: Proceedings of the 4th Asian Conference on Pattern Recognition, pp. 634–639. IAPR, Nanjing, December 2017. https://doi.org/10.1109/ACPR.2017.139
- 11.Ruder, S.: An overview of gradient descent optimization algorithms, June 2017. http://ruder.io/optimizing-gradient-descent/