Indexed by:
Abstract:
Hand tracking is a crucial step in vision based gesture interaction. Due to illumination and hand shape variation, tracking of such object is a difficult task. In this paper, an ensemble tracking framework integrated with superpixels for hand tracking is proposed. With treating tracking as a binary classification problem, ensemble tracking utilizes weak classifiers to get strong classifier and get robustness to illumination and appearance change. However, outliers in confidence map caused by irregular object shape bring ill effects to tracking performance. As hand area has similar color and texture, superpixel method can be used to cluster the pixels in hand area and provide solution to exclude outliers. Experiments show that with superpixe integrated, the proposed method get more accurate tracking result. © Springer International Publishing Switzerland 2013.
Keyword:
Reprint Author's Address:
Email:
Source :
ISSN: 0302-9743
Year: 2013
Volume: 8294 LNCS
Page: 690-697
Language: English
Cited Count:
SCOPUS Cited Count:
ESI Highly Cited Papers on the List: 0 Unfold All
WanFang Cited Count:
Chinese Cited Count:
30 Days PV: 2