Open AccessSpatial Histograms for Region‐Based Tracking
Author(s) -
Birchfield Stanley T.,
Rangarajan Sriram
Publication year - 2007
Publication title -
etri journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.295
H-Index - 46
eISSN - 2233-7326
pISSN - 1225-6463
DOI - 10.4218/etrij.07.0207.0017
Subject(s) - histogram , artificial intelligence , weighting , similarity (geometry) , color histogram , pattern recognition (psychology) , tracking (education) , mean shift , similarity measure , measure (data warehouse) , histogram matching , computer science , intersection (aeronautics) , spatial analysis , computer vision , histogram equalization , particle filter , mathematics , data mining , geography , image (mathematics) , image processing , statistics , color image , cartography , medicine , psychology , pedagogy , radiology , kalman filter
Spatiograms are histograms augmented with spatial means and covariances to capture a richer description of the target. We present a particle filtering framework for region‐based tracking using spatiograms. Unlike mean shift, the framework allows for non‐differentiable similarity measures to compare two spatiograms; we present one such similarity measure, a combination of a recent weighting scheme and histogram intersection. Experimental results show improved performance with the new measure as well as the importance of global spatial information for tracking. The performance of spatiograms is compared with color histograms and several texture histogram methods.