Many approaches for retrieving the images semantically have been proposed. There still have
many known limitations. The key problem is the identification of appropriate concepts. In this
paper, we introduce the new image retrieval system that employs a concept based technique
utilizing ontology. There are many attempts to search images using ontology. However, they
haven’t given the much good results. The reason is that the ontology just has been used to
resolve the conceptual heterogeneous between the text annotations. Another reason is to use the
much big ontology. To improve the accuracy in terms of precision and recall of an image retrieval
system we have created a personalized ontology and spatial ontology. In trial implementation
of our system we have achieved a level of accuracy at which was up to 83.9%.