Abstract
Due to the advance in computer and network technology, the amount of multimedia data, such as images and sounds, makes organizing them an arduous task. How do we deal with this enormous amount of data? We preprocess the data with computer to increase time efficiency and save manpower. We analyze images and make automated annotations via previously built ontology. The past image retrieval systems utilize low-level features such as color, texture and shape, to analyze images, but this does not take semantics of content and physical objects into account which usually leads to misunderstanding of images and false indexing, recognition, and annotation. We propose a method that applies semantics of content on analysis of physical objects in images, so computers can accurately detect objects, deduce the relations between objects and extract the underlying semantics. With sufficient annotations, users can query desired images with natural language in the future. Comparing between conventional low-level annotations and our newly proposed ontology-based annotations, ontology enhances the comprehension of images by computers, and the accuracy of object recognition is also increased. We will demonstrate the ability of keywords annotation, region naming, and descriptively commenting in our system.