Algorithm for real-time finger spelling alphabet recognition in video sequences
Algorithm for real-time finger spelling alphabet recognition in video sequences
复制标题
视频序列中的实时手指拼写字母识别算法
DOI：
10.1109/sps.2017.8053688
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
J. Polec
中科院分区：
文献类型：
--
作者：
Filip Csoka;J. Polec
本文主要研究视频序列中的真实的=时间图像识别问题。沿着介绍了一种专门用于手指拼写字母识别的新算法，并与现有技术进行了定性比较。提出的概念引入了一种新的图像特征选择技术，用于比较过程，显著提高了视频序列中手指拼写字母的识别率。考虑到现代技术能够记录视频和捕获具有非常高分辨率的图像的事实，并且所提出的新算法完全能够有效地处理高分辨率图像。
This paper focuses on real=time image recognition in video sequences A novel algorithm designed to be used specifically for finger spelling alphabet is introduced along with relevant qualitative comparison to current state of the art. The proposed concept introduces a new image feature selection technique used in comparation process, which significantly improves the recognition of finger spelling alphabets in video sequences. Consideration to the fact that modern technologies are capable of recording videos and capturing images with very high resolution was taken and the proposed novel algorithm is fully capable of efficiently processing also high resolution images.