Development of a computer vision-based technique for analysis of hand therapy exercise
Hand therapy exercises based on computer vision system have many benefits and this has attracted the interest of researchers towards building a computer vision application. The innovation of measurement tool in computer vision application for the exercise also have undergone continuously improvement...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/26115/1/Development%20of%20a%20computer%20vision-based%20technique%20for%20analysis.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Hand therapy exercises based on computer vision system have many benefits and this has attracted the interest of researchers towards building a computer vision application. The innovation of measurement tool in computer vision application for the exercise also have undergone continuously improvement which is using wearable sensor, robot aid and computer vision. The hand therapy exercise system that was built using computer vision method involved highly flexible structure of the human body and causes a self-occlusion. Another problem is tracking techniques using an ordinary camera are not easy and require extensive time to develop. The aim of this research is to develop a computer vision-based technique for analysis of hand therapy exercise and to evaluate the propose angle measurement with the conventional measurement tool, which is goniometer. In this thesis a Range of Motion (ROM) exercises for wrist and elbow by using the Kinect sensor is presented and this exercise provides a real-time feedback for every movement. Both exercises, which are wrist radial/ulnar deviation and elbow flexion/extension also consists of joint angle measurement. For this purpose, the ROM exercise was developed by using the hand detection and tracking, fingertips detection, skeleton joints detection algorithms and joint angle measurement algorithm. Then, the joint angle values can be saved automatically. The result for the measurement of wrist radial/ulnar deviation angle shows its value is different from the reference angle of ±6°. Meanwhile, the result for the measurement of elbow flexion/extension joint angle was different at ±8° from the reference angle. These measurement values are almost similar to the standard ROM reference values which are 20° for radial deviation, -30° for ulnar deviation, 140° for flexion and 0° for extension. The algorithm worked well and could follow the movement of the upper arm and forearm in real-time. This exercise system that uses Kinect sensor is able to be a portable exercise tool, which is easy to install and to be placed anywhere. The joint angle measurement data can be saved and can be used for reference in the future. The validation with the conventional tool for angle measurement (goniometer) shows that the joint angle measurement with the proposed technique is more precise compared to the goniometer because it can measure the small scale of angle in degree. |
|---|