| dc.description.abstract | In recent years, wheelchair is the assistive device for the disabled or the elderly. However, traditional wheelchair, including powered and manual wheelchair, is still highly unsuitable especially for severely disabled. In this research, biomedical image processing technique is applied to wheelchair to help the severely disabled move easier. In biomedical image processing, facing detection is a potential method to detect behavior and movement of face to control the motion of wheelchair. In this research, movement of head (or face – inclination) is utilized to control direction of wheelchair (left, right or center) and the shape of mouth is used to control two status of this one. By combining Adaboost algorithm, Haar feature to detect whole face. After that, principle component analysis (PCA) theory is applied to verify mouth state. (open or close)
A Labview program was created to control an electric wheelchair by processing motion and behavior of face’s user. In the thesis, the analysis face behavior program has two functions. First of all, it is built to detect face and then follows movement of head by using and Principle Component Analysis (PCA) is employed to detect the state of mouth (open or close). After calculating the average of the sum of whole value pixels in region of interested of the mouth, the threshold is created. This threshold decides the status of the mouth. Thus, the accuracy will be increased by increasing the number of samples. The experimental results obtained in this thesis are performed in indoor environment in laboratory of Bio Medical Engineering Department at International University. The result can be applied not only in controlling electric wheelchair but also in another devices for supporting the severely handicap or the elderly in daily life.
Keyword: Smart wheelchair, Haar features, Adaboost algorithm, Facing detection, Principle component analysis (PCA). | en_US |
