Automatic Ontology Derivation Using Clustering for Image Classification.
Latifur Khan, Lei Wang · 2002
Technology in the field of digital media generates huge amounts of non-textual information, audio, video, and images, along with more familiar textual information. The potential for exchange and retrieval of information is vast and daunting. The key problem in achieving efficient and user-friendly retrieval in the domain of image is the development of a search mechanism to guarantee delivery of minimal irrelevant information (high precision) while insuring that relevant information is not overlooked (high recall). The traditional solution to the problem of image retrieval employs contentbased search techniques based on color, texture or shape features. The traditional solution works well in performing searches in which the user specifies images containing a sample object, or a sample textural pattern, in which the object or pattern is indexed. One can overcome this restriction by indexing images according to meanings rather than objects that appear in images, although this will entail a way of converting objects to meanings. We have solved this problem of creating a meaning based index structure through the design and implementation of a concept-based model using domain dependent ontologies. An ontology is a collection of concepts and their interrelationships which provide an abstract view of an application domain. With regard to converting objects to meaning the key issue is to identify appropriate concepts that both describe and identify images. We propose a new mechanism that can generate ontologies automatically in order to make our approach scalable. To achieve this we propose a method for the automatic construction of ontologies based on clustering and a vector space model. Similarity of images is based on similarity of objects that appear in images. For object similarity measure, we consider the combination of color and shape similarity together. 1.