Research On Ontology-based Image Affective Retrieval

With the development of the computer technology, network technology, emetic network technology and the multimedia technology, come along with fast development of the man to computer system, More and more people focus their research on the emotion which in the virtual feature, especially in the area of retrieval feature and the area of classed images. At present, most dealed pictures and used pictures was ignoring the emotion effects. Make an effort to discover the relationship between picture and emotional was the key and the difficult during the studying retrieve emotion images, but it was a new and challenge research subject.Images not only contain visual features, but also contain words information and emotion information. At present, the popularity image retrieval system is content-based image retrieval which is based on visual features (color, shape and texture).It lack"emotion"and"effective"during the retrieval, so the result is not perfect, if we can make picture connect with emotion, it will contribute to mutual communicates between people and machine, and the subjects based on emotion computer and so on.In this paper, firstly, analyse the basic aknowledge of ontology,MPEG-7 and classed emotion, then the used the ontology theory, based on the description schema of the MPEG-7,turn the images feature data which descript by XML to descript by RDF(resource description frame) and OWL(Web Ontology Language) ,turn the image features data into the data which descript by OWL, used RDF and OWL to organize emotion semantic of the picture,build a picture emotion semantic mode which based on ontology, according the image physic informations to descript emotion information, according to the emotion sorts, according Jena reason rules syntax and OWL reason rules to write useful semantic rules, build reasoner engine. Realize mapping from picture physics features to emotion space, and then carry out emotion based image retrieval. In my experience, I used protégé3.4 to build ontology documents, used Jena2.5.7 to parse OWL data, used eclipse as exploit environment, build reason rules, realized myself define reasoner,then validate the method which the paper mention.