基于HMM_SVM的中国手语识别研究  

Study on HMM_SVM-based Chinese Sign Language Recognition

作　　者：赵蔚宁 

机构地区：[1]吉林省福彩中心协会筹建办公室,长春130061

出　　处：《长春大学学报》2011年第10期24-26,共3页Journal of Changchun University

摘　　要：提出了基于隐马尔可夫方法的中国手语识别方法,通过对于手势的取像,图像处理,降维技术,通过应用Sugeno模糊积分,图像处理技术是计算机图形学识别的基本步骤,采用直方图的特征可以将手型的区域从背景中分离出来,然后通过降维处理,将得到的手部图像去除手部以外的区域,从而得到手的轮廓;使得不携带其他相关手套工具,达到静态简单手语识别,其正确率达到85%以上,证明这种方法在中国手语识别上的可行性。This paper presents a Hidden Markov-based Chinese sign language recognition method by means of image capture of gestures,image processing and dimensionality reduction techniques.Through the application of Sugeno fuzzy integral,image processing technology is basic step for computer graphics identification,which can separate the hand-shaped area from the background by using histograms features,and then removes the part outside hand from obtained hand image by dimension reduction process to get the profile of hand.This method realizes the static identification for simple sign language without related gloves tools,the correct rate reaches 85% or more and the feasibility of sign language recognition in China is proved.

关 键 词：多维分割 隐马尔科夫算法 SVM(support VECTOR machine支持向量机) 中国手语识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]