Small-Variance Nonparametric Clustering on the Hypersphere
This work addresses clustering challenges in computer vision and other domains with directional data, offering incremental improvements over existing methods.
The authors tackled the problem of clustering directional data like surface normals by proposing two new k-means-like algorithms based on the small-variance limit of Bayesian nonparametric von-Mises-Fisher mixtures, achieving efficient performance on synthetic and real 3D data while generalizing to high-dimensional applications.
Structural regularities in man-made environments reflect in the distribution of their surface normals. Describing these surface normal distributions is important in many computer vision applications, such as scene understanding, plane segmentation, and regularization of 3D reconstructions. Based on the small-variance limit of Bayesian nonparametric von-Mises-Fisher (vMF) mixture distributions, we propose two new flexible and efficient k-means-like clustering algorithms for directional data such as surface normals. The first, DP-vMF-means, is a batch clustering algorithm derived from the Dirichlet process (DP) vMF mixture. Recognizing the sequential nature of data collection in many applications, we extend this algorithm to DDP-vMF-means, which infers temporally evolving cluster structure from streaming data. Both algorithms naturally respect the geometry of directional data, which lies on the unit sphere. We demonstrate their performance on synthetic directional data and real 3D surface normals from RGB-D sensors. While our experiments focus on 3D data, both algorithms generalize to high dimensional directional data such as protein backbone configurations and semantic word vectors.