Developing an assistive technology to help children with autism for recognising human emotion

Assistive Technology (AT) is a means of product system that is used to increase, maintain or improve functional capabilities of individuals and disabilities. AT promotes greater independence by enabling disabled people to perform task that they were formerly unable to accomplish, by providing enhancements to or changed methods of interacting with the technology needed to accomplish such tasks. Children with autism spectrum disordered (ASD) have difficulty recognising emotions in themselves and others. This work presents a portable device with the capability of recognising human emotional state which is part of the AT designed to help children with ASD to read and respond to the facial expressions of people they interact with. An input image is obtained via a camera embedded with the proposed device and subsequently Gabor feature extraction is used to extract features from the image. The underlying algorithm adopted Gini index for feature reduction and Naïve Bayesian Classifier (NBC) is proposed to determine the corresponding facial emotion. The proposed device is able to provide an accuracy of 75% for face expression recognition performing under a Windows mobile platform. The system training time is around five seconds and the processing speed is around two frames per second. Our future research direction includes the extension of this work to be deployed in Symbian and iPhone operating system