3D model based gesture acquisition using a single camera

被引:5
作者
Horain, P [1 ]
Bomb, M [1 ]
机构
[1] Inst Natl Telecommun, Dept EPH, Grp Ecole Telecommun, F-91011 Evry, France
来源
SIXTH IEEE WORKSHOP ON APPLICATIONS OF COMPUTER VISION, PROCEEDINGS | 2002年
关键词
D O I
10.1109/ACV.2002.1182175
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We present a method for 3D human motion capture using a single camera, without markers and without a priori knowledge on gestures. It is based on registering a 3D articulated model on color images with respect to biomechanical constraints. Gestures regularization is discussed as a way to cope with projection ambiguities. Computation is reduced by registering only the moving parts of the body.
引用
收藏
页码:158 / 162
页数:5
相关论文
共 24 条
[1]   Robustly estimating changes in image appearance [J].
Black, MJ ;
Fleet, DJ ;
Yacoob, Y .
COMPUTER VISION AND IMAGE UNDERSTANDING, 2000, 78 (01) :8-31
[2]  
BOBICK AF, 1997, COMP IMAG VIS, V9, P125
[3]   Discovery and segmentation of activities in video [J].
Brand, M ;
Kettnaker, V .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2000, 22 (08) :844-851
[4]  
CUTLER R, 1998, P IEEE FG 98
[5]  
DAVIS ER, 1995, MACHINE VISION THEOR
[6]   3D articulated models and multiview tracking with physical forces [J].
Delamarre, Q ;
Faugeras, O .
COMPUTER VISION AND IMAGE UNDERSTANDING, 2001, 81 (03) :328-357
[7]  
GVRILA DM, 1996, P IEEE CVPR SAN FRAN, P73
[8]  
HORAIN P, 2002, ACQUISITION GESTE VI
[9]  
IIYAMA M, 2000, P ICPR 00, V3, P699
[10]  
IWAI Y, 1997, P IROS 97 SEPT, V2, P960