Content-based image retrieval using fuzzy visual representation

In this paper, a fuzzy approach is used for efficiently representing the visual content. Initially, the image is partitioned into several segments (objects) and for each segment appropriate features, such as the segment color and texture, are extracted. In the following, all features are classified in a fuzzy framework, resulting in a content interpretation closer to the human perception. Furthermore, such fuzzy representation of visual content gives to the users the capability of expressing video queries using statements drawn from the natural language. An objective criterion has been also proposed to evaluate the system reliability, which indicates the advantages of the proposed fuzzy approach compared to other traditional techniques for image retrieval.