The Research Of Dynamic Sign Language Recognition Based On Video

With the rapid popularization of the computer, a new generation of human-centered human-computer interaction technology has been becoming the current hot issues. Gesture is a kind of natural and intuitive means of communication between human and computer, so gesture recognition is one of the key technologies to realize the new generation of human-computer interaction. At the same time, sign language is the method for deaf people to communicate and exchange ideas instead of language through finger alphabet and gestures. Most of the hearing know little about sign language, so it is hard for them to communicate with the deaf. And as a new occupation in China, sign language interpreting is difficult to meet the demand. Therefore, research on sign language recognition technology can make the communication between deaf and hearing people more convenient and accessible. What’s more, it can also promote the development of the computer related fields.A new method for sign language recognition is proposed in this paper, which is based on key frame extraction. The sign language can be recognized in real-time by reducing the processing data through turning the video into gestures sequence. To achieve the key frame extraction, an adaptive algorithm which can adapt to different signers, different sampling frequency is proposed in this paper. The key frames can be regarded as sign language units and the different combinations of the units represent the sign languages. But different units show different importance for different sign language. According to this, an algorithm for recognition based on weighted units is proposed. The weights are measured by different importance. Then the random forest is used to classify the units and recognize the combination as a sign language finally. The test shows that the algorithm proposed in this paper can both extract the key frame and achieve the recognition effectively in real-time.