An algorithm for data-driven bandwidth selection
Top Cited Papers
- 19 February 2003
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 25 (2) , 281-288
- https://doi.org/10.1109/tpami.2003.1177159
Abstract
The analysis of a feature space that exhibits multiscale patterns often requires kernel estimation techniques with locally adaptive bandwidths, such as the variable-bandwidth mean shift. Proper selection of the kernel bandwidth is, however, a critical step for superior space analysis and partitioning. This paper presents a mean shift-based approach for local bandwidth selection in the multimodal, multivariate case. The method is based on a fundamental property of normal distributions regarding the bias of the normalized density gradient. This paper demonstrates that, within the large sample approximation, the local covariance is estimated by the matrix that maximizes the magnitude of the normalized mean shift vector. Using this property, the paper develops a reliable algorithm which takes into account the stability of local bandwidth estimates across scales. The validity of the theoretical results is proven in various space partitioning experiments involving the variable-bandwidth mean shift.Keywords
This publication has 23 references indexed in Scilit:
- Finding Salient Regions in ImagesComputer Vision and Image Understanding, 1999
- Distribution Free Decomposition of Multivariate DataPattern Analysis and Applications, 1999
- A Brief Survey of Bandwidth Selection for Density EstimationJournal of the American Statistical Association, 1996
- A Brief Survey of Bandwidth Selection for Density EstimationJournal of the American Statistical Association, 1996
- A transform for multiscale image segmentation by integrated edge and region detectionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1996
- Smoothing Bias in Density Derivative EstimationJournal of the American Statistical Association, 1993
- Divergence measures based on the Shannon entropyIEEE Transactions on Information Theory, 1991
- Comparison of Data-Driven Bandwidth SelectorsJournal of the American Statistical Association, 1990
- Comparison of Data-Driven Bandwidth SelectorsJournal of the American Statistical Association, 1990
- On Bandwidth Variation in Kernel Estimates-A Square Root LawThe Annals of Statistics, 1982