Loading...

Combined Wrist and Forearm Movement Recognition using sEMG

Karbasi, Hamed | 2018

417 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 51451 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Jahed, Mehran
  7. Abstract:
  8. Physiotherapy is a major part of the rehabilitation process that is used to retrieve patients' physical ability. The recuperation feedback in the physiotherapy process has a twofold significance for both physiotherapists and patients. It helps patients regain their ability to recover more quickly, prevent recurrence of injury to the treated area and other areas, and motivate the patient to continue treatment. It also helps the physiotherapist to monitor the process of rehabilitation and ensure the correctness of the procedure. Meanwhile, many patients cannot maintain continuous workouts due to lack of access to the physiotherapist at home and inability to provide required feedback.In this study, a system based on the electromyogram signal and computer vision is presented, which performs classification between the three classes of continuous forearm movement, contraction and expansion of the wrist in combined movements. The method of "Bag of Word" is used for feature extraction and voting between two support vector machines (SVM) of the electromyogram signal and computer vision is used for classification. Furthermore, if the forearm movement is detected, the angle of the movement is estimated using multilayer perceptron neural network. The accuracy of classification based on time-dependent spectral Features was 97.8%. In the classification part, the convergence analysis shows that this system requires an average of 4.8 seconds to classify the movement. Also, in the regression section to estimate the angle, accuracy of 87% was obtained
  9. Keywords:
  10. Words Bag Model ; Feature Extraction ; Support Vector Machine (SVM) ; Neural Networks ; Regression Analysis ; Surface Electromyogram

 Digital Object List

 Bookmark

No TOC