Title :
Finger contour tracking based on model
Author :
Zhou, Hang ; Ruan, Qiuqi
Author_Institution :
Inst. of Inf. Sci., Northern Jiaotong Univ., Beijing, China
Abstract :
In this paper, we study the articulated hand model with a set of transforms featured with angles among the fingers. The feature points are extracted from images of the unmarked hand from a viewpoint. With the features extracted from the images, the system calculates them for each line and tip in the model with contour correction. Then the modified Gauss-Newton algorithm minimizes the feature residuals. The projection of the previous estimated model is used to hypothesise that the closest available feature is the correct match.
Keywords :
Newton method; computer vision; edge detection; feature extraction; image matching; image sampling; importance sampling; minimisation; tracking; DoF; articulated hand model; contour correction; correct match; feature point extraction; feature residual minimization; finger contour tracking; modified Gauss-Newton algorithm; sampling; tracking; transforms; vision based recognition system; Data mining; Electronics packaging; Feature extraction; Fingers; Information science; Kinematics; Least squares methods; Newton method; Thumb; Wrist;
Conference_Titel :
TENCON '02. Proceedings. 2002 IEEE Region 10 Conference on Computers, Communications, Control and Power Engineering
Print_ISBN :
0-7803-7490-8
DOI :
10.1109/TENCON.2002.1181323
