作者: Alireza Asvadi , Hami Mahdavinataj , Mohammadreza Karami , Yasser Baleghi
DOI: 10.1007/978-3-319-10849-0_8
关键词:
摘要: This paper presents an object tracking algorithm based on discriminative 3D joint RGB histograms of the and surrounding background. Mean-shift confident map is used for localization. An incremental color learning scheme with a forgetting factor utilized to account appearance variation object. Evaluated against three state art methods, experiments demonstrate effectiveness proposed where undergoes in illumination color. Implemented MATLAB, tracker runs at 25.7 frames per second.