Mechanomyogram (MMG)-based pattern recognition refers to the process of collecting MMG bands and applying machine learning algorithms to perform motion recognition. To realize the real-time classification of sign language motions, six-channel MMG signals of three muscles on forearm are collected by dual-axis acceleration sensors which are controlled by lightweight embedded device with STM32 chip. The back propagation neural network (BPNN) algorithm is used to establish recognition classification models, where the parameters are extracted and put into the embedded system to transfer BPNN algorithm. The embedded device can accomplish real-time recognition of 30 kinds of sign language motions, with the self-test accuracy up to 99.6% and the accuracy of real-time recognition up to 97.5%. Moreover, the classification time for each motion is less than 0.52 ms, satisfying the real-time recognition condition. The results can be applied to the fields of rehabilitation engineering, sign language translator, and prosthetic control, etc.
表 1 模型自测结果Table 1 Self-test results
表 3 多种分类器自测识别率对比Table 3 Self-test of different classifiers
图1 技术路线流程图Fig.1 Flow chart of technical process
图2 常见的采集形式Fig.2 Commonly-used data collector
图3 环形弹性带采集装置Fig.3 Elastic bracelet data acquisition device
图4 金属表带结构采集装置Fig.4 Metal braclet data acquisition device
图5 基于指数加权平均的能量阈值分割算法Fig.5 Energy threshold segmentation algorithm based on exponential weighted average
图6 NN模型框架(3层隐含层,神经元个数分别为80,50,30)Fig.6 NN architecture (three hidden layers with neuron numbers of 80, 50 and 30)
图7 采集手语动作样本Fig.7 Collection of sign language motion data
图8 传感器对应的前臂三块肌肉Fig.8 Three muscles of the forearm corresponding to sensors
图9 30种中国标准手势语言Fig.9 30 kinds of Chinese sign language gestures
图10 志愿者1实时测试结果混淆矩阵Fig.10 Confusion matrix of volunteer 1