AN AGENT-BASED IMAGE DESCRIPTION FRAMEWORK

Document Type : Original Article

Authors

College of Computer Science and Engineering, King Fahd University of Petroleum and Minerals

Abstract

This paper presents high quality graphical computer software that helps the visually impaired people to interact with the graphical computer software. We note that almost every document contains some kind of images which make a problem for special users like visually impaired people to interact and understand the context of the document. As we know that there is a support for text to be converted into voices but there exist no support for an image to be converted into a voice. This issue has motivated us to develop a framework that can give a text description of the images, where its real description significantly aids the work of both visually impaired and professionals. Based on this idea, we have developed an Agent-based Image Description Framework (ABIDF) that enables visually impaired people to recognize most common geometrical shapes in addition to some normal images. The framework produces phrases to describe the recognized shapes, and finally submits the generated text description to the voice synthesizer to be converted into speeches. This process will help the visually impaired users to identify these charts and make the required information more available to them. To support the usefulness of our work, experimental applications are introduced and the obtained converted images into voices showed successful and encouraging results of the proposed framework.

Keywords