Mean shift: a robust approach toward feature space analysis Export

@article{citeulike:462300, abstract = {A general non-parametric technique is proposed for the analysis of a complex multimodal feature space and to delineate arbitrarily shaped clusters in it. The basic computational module of the technique is an old pattern recognition procedure: the mean shift. For discrete data, we prove the convergence of a recursive mean shift procedure to the nearest stationary point of the underlying density function and, thus, its utility in detecting the modes of the density. The relation of the mean shift procedure to the Nadaraya-Watson estimator from kernel regression and the robust M-estimators; of location is also established. Algorithms for two low-level vision tasks discontinuity-preserving smoothing and image segmentation - are described as applications. In these algorithms, the only user-set parameter is the resolution of the analysis, and either gray-level or color images are accepted as input. Extensive experimental results illustrate their excellent performance}, address = {Los Alamitos, CA, USA}, author = {Comaniciu, D. and Meer, P.}, citeulike-article-id = {462300}, citeulike-linkout-0 = {http://doi.ieeecomputersociety.org/10.1109/34.1000236}, citeulike-linkout-1 = {http://dx.doi.org/10.1109/34.1000236}, citeulike-linkout-2 = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1000236}, day = {07}, doi = {10.1109/34.1000236}, issn = {0162-8828}, journal = {Pattern Analysis and Machine Intelligence, IEEE Transactions on}, keywords = {feature, filtering, mean-shift, robust, segmentation}, month = {August}, number = {5}, pages = {603--619}, posted-at = {2007-11-28 16:20:55}, priority = {5}, publisher = {IEEE Computer Society}, title = {Mean shift: a robust approach toward feature space analysis}, url = {http://dx.doi.org/10.1109/34.1000236}, volume = {24}, year = {2002} }

TY - JOUR ID - citeulike:462300 L3 - citeulike-article-id:462300 N2 - A general non-parametric technique is proposed for the analysis of a complex multimodal feature space and to delineate arbitrarily shaped clusters in it. The basic computational module of the technique is an old pattern recognition procedure: the mean shift. For discrete data, we prove the convergence of a recursive mean shift procedure to the nearest stationary point of the underlying density function and, thus, its utility in detecting the modes of the density. The relation of the mean shift procedure to the Nadaraya-Watson estimator from kernel regression and the robust M-estimators; of location is also established. Algorithms for two low-level vision tasks discontinuity-preserving smoothing and image segmentation - are described as applications. In these algorithms, the only user-set parameter is the resolution of the analysis, and either gray-level or color images are accepted as input. Extensive experimental results illustrate their excellent performance IS - 5 JF - Pattern Analysis and Machine Intelligence, IEEE Transactions on SN - 0162-8828 AD - Los Alamitos, CA, USA EP - 619 TI - Mean shift: a robust approach toward feature space analysis PB - IEEE Computer Society VL - 24 SP - 603 KW - feature KW - filtering KW - mean-shift KW - robust KW - segmentation AU - Comaniciu, D AU - Meer, P PY - 2002/08/07/ UR - http://dx.doi.org/10.1109/34.1000236 ER -