Classification of hand movements related to grasp by using EMG signals

This study is aimed classification of Electromyography signals associated with the six hand movements which can be considered as daily hand grasps. In first stage of the study, histogram values of all EMG signals were calculated. Then, by calculating the average value of the histogram values for each of six hand gestures, common features of the movement was determined. Separability into only two or three classes of data have been identified when these average values were examined. Therefore the histogram data were previously divided into three classes. Then each of these classes were divided into two classes. As a result six different classes were created. On the other hand the same cascade classification process was evaluated by dividing three classes of each of previously obtained two classes. When two different cascade classification results were compared it was determined that the 100% success rate can be achieved by dividing three classes into previously obtained two classes. As a result, for the classification of electromyography signals associated with hand grasp movement it has been proposed a very successful and efficient feature extraction and cascade classification method.