Research On Chinese Static Sign Language Recognition

With the rapid development of science and technology,interaction between human beings and computers has become an important part of our daily life.One of the most important issues of the interaction is the gesture recognition.Sign language recognition,the most similar one to the gesture recognition,is to through computer provide an efficient and accurate mechanism of translating sign language into texts or speech sounds in order to realize communication between natural language and sign language,thus helping the deaf communicate with the able-bodied in a more convenient and efficient way so that they can integrate into society better.In this paper,Chinese sign language recognition based on static image processing technique is studied in the following aspects:establishment of database,pretreatment of images,and feature extraction.All the image data in this paper are either made with digital camera or collected from websites of Chinese Special Education and National Association of the Deaf,and some CDs for education.Image-based sign language is pretreated,including image smoothing and denoising,grayness,and edge detection,and then is recognized by adopting the Fourier coefficient as characteristic vector in feature extraction,because Fourier Description of the image is of a good rotation,size,and translational invariance.In order to reduce errors in recognition similar gestures,and increase recognition rate,in this paper based on abundant experiments a threshold "r" is added to Euclid Distance in the classification of pictures of sign language.The experimental results show that when the Fourier coefficient is taken as characteristic vector,Euclid Distance threshold of the recognition rate reaches 96.2%.