基于IMU穿戴式设备的手语识别研究-机械工程专业论文.docxVIP

广东工业大学硕士学位论文ABSTRACT 广东工业大学硕士学位论文 ABSTRACT Gesture recognition is important research field of human—computer interaction and pattern recognition，and the recognition of sign language belonging to gesture recognition．Study of sign language recognition be divided into two ways：visual based and non visual based wearable devices．Based on the analysis of sign language element，this paper made improvement of sign language recognition based wearable devices，proposed skeletal model of sign language，aims to improve the stability and accuracy of sign language recognition and make contributions to the practical application of sign language in real time． The main research of this paper include the following aspects： (1)Analyse the papers related to sign language and sign language recognition based wearable devices．And summarize the difficult problem in this field，including the original data acquisition and processing，sign language model to determine，selection of recognition algorithm and research ideas and experimental scheme to sort out． (2)In order to obtain the motion posture of the hand，this paper uses the IMU sensor to obtain the original data of the hand movement，using the improved complementary filtering algorithm for data fusion to get the Euler angle and convert the Euler angles to quaternions for convenient calculation． (3)Analyze the sign language elements，specifying that the completion of sign language requires the cooperation between hand and upper body．The sign language skeleton model is put forward，and the model is verified by the forward kinematics knowledge of the robot． (4)Use the Slerp interpolation method to make the training data signal of the same dimension，a specific solution is put forward． (5)Training and learning 50 sign language gesture signals by sparse representation and dictionary learning with a high recognition rate． Keywords：IMU；quarternion；sign language skeleton model；dictionary learning Ⅱ 万方数据 目录目录 目录 目录 摘要 I ABSTRACT ．II 目录 ．III Contents ． ．