Lip Expressions Based Control via Image Processing

Abstract

This paper presents an image processing technique to identify lip expressions in producing control instructions. The purpose is to assist patients suffering from Tetraplegia especially in their mobility by providing alternative ways of controlling devices. In this work, motorized wheelchairs are used as the studied mobility device. Some of the methods used in motorized wheelchair do give user difficulties as they need extra effort in controlling the device. Thus, this project aimed to study, design and implement an image processing technique to differentiate four different lips expressions. The classified four different lip expressions will then be used as instructions for forward, turn right, turn left and stop movements of motorized wheelchairs. A technique of morphological processing is applied to the image of the face in order to extract the lips region. The obtained region is then calculated for several parameters. The parameters classification is done using Artificial Neural Network (ANN) to determine the category of lips expressions. In the performance analysis, the developed technique is proven to successfully classify the lips expressions into the categories and produce appropriate instructions for the use in motorized wheelchair control.