|
KMID : 1144120120020020129
|
|
Biomedical Engineering Letters
2012 Volume.2 No. 2 p.129 ~ p.137
|
|
|
Classification of nine directions using the maximum likelihood estimation based on electromyogram of both forearms
|
|
Yu Song-Hyun
Jeong Eui-Chul
Hong Ki-Hwan
Lee Sang-Min
|
|
|
Abstract
|
|
|
Introduction: In this paper, the nine directions are classified using the maximum likelihood estimation (MLE) based on electromyogram (EMG) obtained from both forearm. The authors compared the EMG features to confirm classification accuracy of nine directions.
Methods: Fifteen subjects participated in the experiment, were asked to act three types of motions for each forearm. The motions of left wrist were forward, backward, and rest, while the motions of right wrist were left, right, and rest. Three motions of left wrist and three motion of right wrist were combined to classify the nine directions. The EMG features with 166ms time-window for five seconds of each motion were extracted from the obtained EMG signals. The difference absolute mean value (DAMV), difference absolute standard deviation value (DASDV), mean absolute value (MAV), and root mean square (RMS) were chosen as the EMG feature to compare classification accuracy of nine directions.
Results: In the results, the average classification accuracy of 97.71% was confirmed for each of the nine directions. The DAMV had the highest classification accuracy in the EMG features with 99.26% accuracy. And DASDV had classification accuracy with 99.19%.
Conclusions: These results demonstrate that in this study the classification performance using DAMV or DASDV is stronger than that using MAV or RMS. Furthermore, the classified direction using DAMV or DASDV can be used as precise input data for controlling of robots and devices in rehabilitation applications.
|
|
|
KEYWORD
|
|
|
Electromyogram (EMG), Maximum likelihood estimation (MLE), EMG feature, Both forearms, Nine directions
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|
|