Paper Abstract and Keywords |
Presentation |
2013-12-19 10:25
Improvement of Discrimination Process and User Interface in the Fingerspelling Training System Using Kinect Yoto Tanaka (NTUT), Daisuke Takabayashi (Univ. of Tsukuba), Masaki Kondo, Nobuko Kato (NTUT), Kazuhiro Fukui (Univ. of Tsukuba), Akio Okazaki (NTUT) |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Fingerspelling is one of useful communication methods for deaf and hard-of-hearing. We are developing a fingerspelling training system using image recognition techniques for the purpose of interactive learning using Kinect. In this paper, firstly, we explain a newly improved fingerspelling training system using multiple depth images from Kinect. In the system, the method for resizing image after segmenting hand part from an input depth image, and the feature extraction method are significantly improved. Next, we show evaluation results of fingerspelling training with a dictionary which is generated form data of sign language interpreters. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Kinect / fingerspelling training system / hand shape clipping / feature extracting / user interface / / / |
Reference Info. |
IEICE Tech. Rep. |
Paper # |
|
Date of Issue |
|
ISSN |
|
Download PDF |
|