Abstract
Background : The idea of developing communication assistive device for disable persons who have lost their motor functions and cannot communicate is challenging. Some paralyzed persons are aware of themselves and environments but cannot move their body parts except their eyes. With the use of an assistive device that can detect an eye or mouth status, these people can partially interact with the surrounding environments. Objective : The purpose of this present study was to develop a machine vision based system for a real-time eye or mouth status detection which can be further applied to be used as assistive controlling or communication device for patients. Material and method : This system composed of a web camera connected to a computer running with the developed software to detect the status of eye or mouth. The software prototype was developed in C++ on an MS Windows 7 PC using OpenCV and Dlib libraries and KNN classification technique. Result : The result showed that the accuracy of eye detection was at the highest accuracy (98%) at the middle distance (140cm), whereas the same distance result also observed with the mouth detection (94%). The results from the satisfaction questionnaire demonstrated that all users-healthy volunteers (n=20), dysarthric patients and caregivers (n=20) satisfied with the developed apparatus and software. The satisfaction score in the using of developed equipments was 4.02 (healthy subjects) and 4.21 (patients) from the highest score of 5. The satisfaction score in software using were 3.91, 4.14 (function test); 3.83, 4.09 (usability test) and 4.18, 4.28 (security test) in healthy subjects and patients, respectively. Summary: In conclusion, our developed system can be efficiently used for further development of a communication or controlling assistive device for disable persons
