Research On Image Retrieval Based On Chinese Ontology

With the dramatic increase of image data on network nodes of the Internet, the types of image are even more complicated, each kind of images are becoming richer and richer, this brings a lot of convenience to people. But at the same time, a huge amount of image resources put a burden on image retrieval, how to retrieve multimedia data efficiently such as image has become a hotspot and difficult point at present. To this end, the paper starts with building an image retrieval system based on Chinese ontology, research technologies related to Chinese ontology and image retrieval based on it, researches on this area are very rare, and the research has a certain application and market value.Firstly, the paper has researched technologies related to ontology and image retrieval. In order to use ontology in information retrieval, it needs series of methods, technologies and tools. For this reason, I have researched construction principles, construction methods, representation languages, inference rules and query languages on ontology, what’s more, I have also researched tools for building and reasoning ontology. In image processing technology, I have researched on smoothing, segmentation, segmentation, feature extraction and feature matching for digital image. Then, I have researched the using process of ontology in information retrieval system, they are ontology building, ontology storage, ontology extension, ontology inference and ontology querying. Based on the researches above-mentioned, I have built Chinese ontology. Choose ontology field and scope; conform classes, instances, properties and relations during them; building ontology and test ontology consistency. After that, for Chinese ontology, I have carried on research and experiment on its storage, expansion, inference and query.According to researches on image retrieval based on semantics and ontology, I have raised a image retrieval model based on Chinese ontology, explained each module of the model, researched how to build image feature database as one key point, provided a algorithm for building image feature and integrated formula for computing similarity,then analyse and verify the formula. When using the image retrieval system to retrieve images, it can achieve user’s search condition’s semantization by Chinese ontology built, then begin to retrieve feature vector based on the similarity measure formula which is raised by author for match searches and feature vectors. At last, analyze the system’s retrieval effiency, recall ratio and precision ratio.