Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
6940263 | Pattern Recognition Letters | 2018 | 10 Pages |
Abstract
In this work, we propose a robust wrist point detection algorithm based on geometric features of the binary hand mask. Circular and elliptical shapes are used to approximate the palm region. Next, a wrist point detection method is proposed. The proposed algorithms are tested on HGR1 database wherein 899 hand gesture images are provided. The experimental results prove that the proposed elliptical method is accurate and effective as compared to the other existing methods. Almost 84% (753 out of 899) of the wrist points are detected accurately with an acceptable error (eâ¯<â¯0.5) for ground truth skin mask of HGR1 database. Out of these 753 accurately detected wrist point 480 belong to error bin eâ¯<â¯0.2. Performance of the proposed method is also tested on real-life scenario wherein skin masks are obtained from different skin detection algorithms. The outcomes are compared with the ground truth skin mask of HGR1 database and comparable results are obtained with multi-seed propagation in multi-layer graph method.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Vision and Pattern Recognition
Authors
Gourav Modanwal, Kishor Sarawadekar,