Bimanual Hand Tracking based on AR-KLT
- [Place of publication not identified] : INTECH Open Access Publisher, 2007.
- Physical Description:
- 1 online resource
- Language Note:
- Restrictions on Access:
- Open Access Unrestricted online access
- Our ARKLT method is very useful for tracking and gesture recognition. As mentioned before, the ARKLT method consists of three points for each hand: the shoulder, elbow and hand. Since the model reflects the articulated motion of the human body which is restrained by the each limb's degree of freedom. That is, the possible region for hand movement is restricted in the elongated region of the shoulder and elbow movement. Therefore, the proposed method can devise effective prediction method, which enables to pre-detect crossing hands based on the body structure. In addition, the proposed method applies the KLT features and their regression line so that the body structure can effectively be fitted into the target. Also, the well-fitted KLT line can provide the exact point of a hand; meanwhile most tracking methods provide the broad region of the target. When it comes to practical uses such as gesture recognition, the find location of the target improves to draw accurate outcome, for example, gesture recognition.
View MARC record | catkey: 27424565