Paper Info
Title
A robust and accurate 3d hand posture estimation method for interactive systems
Abstract
In this paper, a new 3D hand posture estimation system using a single camera and 3 interactive systems are introduced. Existing hand gesture recognition systems estimate hand's 3D models based on image features such as contour or skin texture. However, it was difficult to estimate the wrist rotation because the contour and the texture data do not have enough information to distinguish hand's sides. To solve this problem, we propose a new 3D hand posture estimation system that uses data of nail positions. Nail positions are an important factor to recognize hand's sides. Using nail positions, it becomes possible to detect whether the camera is facing palm or dorsum. In addition, nail areas can be robustly extracted from a skin area by a simple image processing technique. Our Proposed system uses a database consists of data-sets of the hand's contour, the nail positions, and finger joint angles. To estimate the hand posture, the system first extracts the hand's contour and the nail positions from the captured image, and searches for a similar data-set from the database. The system then outputs the finger joint angles of the searched data-set. Our experimental results show high accuracy in the hand posture estimation with the wrist rotation.
Year
DOI
Venue
2010
10.1145/1709886.1709963
Tangible and Embedded Interaction
Keywords
Field
DocType
existing hand gesture recognition,hand posture estimation method,wrist rotation,hand posture estimation,proposed system,nail area,hand posture,hand posture estimation system,interactive system,nail position,simple image processing technique,robot,image features
Computer vision,Finger joint,Wrist,Dorsum,Feature (computer vision),Computer science,Image processing,Gesture recognition,Skin texture,Artificial intelligence,Robot
Conference
Citations 
PageRank 
References 
5
1.01
3
Authors
1
Name
Order
Citations
PageRank
Emi Tamaki114113.11