Chinese Static Sign Language Recognition Based On Edge Gradient Direction Histogram

Sign Language Recognition is a communication way between the deaf and the health.The aim of computer Sign Language Recognition is that:sign language will be translated into text or voice,so that healthy people can "understand sign language";on the other hand,the voice will be translated into sign language used gestures to express, so that deaf people can "understand the voice." Thus the realizations of Sign Language Recognition enable the communication way between the deaf and the health to become more convenient.And deaf people can also easily use of television,telephone,movies, networks,computers and other high-tech products.So they can better integration into the normal social life and greatly enhance their quality of life to improve their educational environment.There are two ways of Sign Language Recognition:one is digital glove-based recognition,the other is vision-based recognition.The latter could imitate the human vision more natural than the former.So a vision-based method is used here to do the recognition research.In the part of edge detection,this paper presents an improved method based on Canny edge detection.The use of cubic B-spline filter operator obtains multi-scale wavelet transform,and extracts gesture image edge at different scales.fusing the scale of gesture edge to be a single pixel edge image.In the side of extracting the feature vector in the gesture,this paper presents a histogram of the edge gradient direction method.Experiments results show that the method has the robustness against illumination changes,position translation and scale changes,but sensitive to rotation.This paper presents a method of Euclidean distance cyclic shift to avoid the sensitivity of rotated gesture image.By this method it will limit the gesture rotation angle in range of 30°.In this paper,regarding the use of edge gradient direction histogram feature as vector a gesture,the cycle of shifting away from European-style approach to sign language recognition,gesture feature vector,experiments show that the identification between the fast calculation of a simple,rapid,and recognition rate of 97.6 percent.In this paper,we use the edge gradient direction histogram as a feature vector and we use Euclidean distance cyclic shift for sign language recognition.The experiment result shows that the compare of gesture feature vector is rapid,simple and fast,and recognition rate of 97.6 percent.