A sEMG Signal based Gait Phase Recognition Method selecting Features and Channels Adaptively

摘要

the present invention by using the muscles and feature extraction in all cases be a separate gait data generated the classifier, and generating the classifier as a way to walking step-by-step for each of the subject through the training process accuracy by calculating, as the accuracy is about the highest sorter surface electromyogram signal based on the walking step recognize how to use the gait recognition step, (a) receives the sample signal of the EMG signals of many channels, the classifier classifying the EMG signal generating for each said channel; (B) receiving the test signal of EMG signals of many channels, by testing the classifier obtain the accuracy, the method comprising selection of the classifier according to the accuracy; And, (c) to provide a configuration that includes the step of recognizing a walking step in the selection of classifiers. By a method as described above, by using by selecting the appropriate features and the channel from the plurality of characteristic value adaptively, it is possible to process all the features and channel walking step recognized more quickly than the conventional method using, the accuracy of recognition significantly It can be improved. ;