计算机应用与软件2016，Vol.33Issue(9)：151-153,3.DOI:10.3969/j.issn.1000-386x.2016.09.036
基于低分辨率视频图像的手语识别方法
SIGN LANGUAGE RECOGNITION METHOD BASED ON LOW RESOLUTION VIDEO
摘要
Abstract
In practical environment,there are a lot of recognition requirements in regard to low resolution sign language video,but only rela-tively limited discrimination information are contained,thus the recognition efficiency is low.In light of this,we proposed a sign language rec-ognition method.Based on the extraction of target area using real-time skin colour feature,the method calculates two recognition feature values of centroid and boundary chain code of target area,uses dynamic time warping algorithm to recognise the starting frame and ending frame of hand gestures in turn,and restores the sign language words in combination with recognition results.In experiment using the public sign lan-guage dataset of the University of South Florida,the use of the proposed method has been compared with current method.The correctly recog-nised sign language words were increased by 21,and the wrong words was decreased by 1.It eliminated the interference of crippled sign lan-guage words,and this proved the effectiveness of the method.关键词
手语识别/动态时间规整算法/数字图像处理Key words
Sign language recognition/Dynamic time warping/Digital image processing分类
信息技术与安全科学引用本文复制引用
严焰,刘蓉..基于低分辨率视频图像的手语识别方法[J].计算机应用与软件,2016,33(9):151-153,3.基金项目
国家社会科学基金项目（12BTQ038）。 （）
