当前位置: X-MOL 学术IEEE Trans. Hum. Mach. Syst. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Implementation of a Character Recognition System Based on Finger-Joint Tracking Using a Depth Camera
IEEE Transactions on Human-Machine Systems ( IF 3.6 ) Pub Date : 2021-04-14 , DOI: 10.1109/thms.2021.3066854
Md. Shahinur Alam , Ki-Chul Kwon , Nam Kim

The joint tracking-based writing system refers to writing characters by changing the position of the finger joint. It is a new research field in interaction-based input systems. However, joint tracking is a very challenging task. In this article, we present a new method for a finger-joint tracking-based character recognition system using a 3-D camera. The proposed method tracks the finger-joint from 3-D information to identify a numerical digit, alphabet, character, special key, or symbol using the distance between the thumb tip and another finger-joint. The recognition is based on Euclidean distance thresholding and geometric slope techniques. Joint data are stored in a 3-D matrix to assign the 3-D coordinate values. The exact character is identified according to the specified definitions. First, a single hand-based digit recognition method is introduced, in which the left or right hand is used. Second, a double hand-based writing system is presented in which both hands are used simultaneously; this system features a full keyboard with 124 different characters. Our results show an overall accuracy of 91.95% and 91.85% for single-hand and double-hand recognition, respectively; with a recognition time of less than 60 ms for each character. An important contribution of this article is that the proposed system can work in both light and dark environments, requires only a small computation area and has a large number of character sets (124 characters). In addition, a region-based user study has been conducted to verify the approach.

中文翻译:

基于深度相机的手指跟踪的字符识别系统的实现

基于关节跟踪的书写系统是指通过更改手指关节的位置来书写字符。这是基于交互的输入系统的新研究领域。但是,联合跟踪是一项非常具有挑战性的任务。在本文中,我们提出了一种使用3-D相机的基于手指关节跟踪的字符识别系统的新方法。所提出的方法使用拇指尖和另一个手指关节之间的距离来跟踪3-D信息中的手指关节,以识别数字,字母,字符,特殊键或符号。该识别基于欧氏距离阈值和几何斜率技术。关节数据存储在3-D矩阵中以分配3-D坐标值。确切的字符根据指定的定义进行标识。第一的,引入了一种基于单手的数字识别方法,其中使用了左手或右手。其次,提出了一种基于双手的书写系统,其中两只手同时使用。该系统具有一个具有124个不同字符的完整键盘。我们的结果表明,单手识别和双手识别的总体准确度分别为91.95%和91.85%;每个字符的识别时间少于60毫秒。本文的重要贡献在于,所提出的系统可以在明亮和黑暗的环境中工作,只需要很小的计算区域,并具有大量的字符集(124个字符)。此外,已经进行了基于区域的用户研究,以验证该方法。提出了一种基于双手的书写系统,其中双手同时使用。该系统具有一个具有124个不同字符的完整键盘。我们的结果表明,单手识别和双手识别的总体准确度分别为91.95%和91.85%;每个字符的识别时间少于60毫秒。本文的重要贡献在于,所提出的系统可以在明亮和黑暗的环境中工作,只需要很小的计算区域,并具有大量的字符集(124个字符)。此外,已经进行了基于区域的用户研究,以验证该方法。提出了一种基于双手的书写系统,其中同时使用双手。该系统具有一个具有124个不同字符的完整键盘。我们的结果表明,单手识别和双手识别的总体准确度分别为91.95%和91.85%;每个字符的识别时间少于60毫秒。本文的重要贡献在于,所提出的系统可以在明亮和黑暗的环境中工作,只需要很小的计算区域,并具有大量的字符集(124个字符)。此外,已经进行了基于区域的用户研究,以验证该方法。每个字符的识别时间少于60毫秒。本文的重要贡献在于,所提出的系统可以在明亮和黑暗的环境中工作,只需要很小的计算区域,并具有大量的字符集(124个字符)。此外,已经进行了基于区域的用户研究,以验证该方法。每个字符的识别时间少于60毫秒。本文的重要贡献在于,所提出的系统可以在明亮和黑暗的环境中工作,只需要很小的计算区域,并具有大量的字符集(124个字符)。此外,已经进行了基于区域的用户研究,以验证该方法。
更新日期:2021-05-25
down
wechat
bug