An Ontology-based Semantic Image Retrieval Research

With the rapid development of computer and multimedia information, digital image is growing quickly. The people are very difficult to quickly mine helpful image from such a tremendous amount of information. When people are seeking solutions to the problem, scholars have found that the "ontology" will not only help to improve the sharing of information, but also improve the accuracy of information retrieval. This paper attempts to connect image retrieval with ontology knowledge, puts forward a new method of semantic image retrieval based on ontology.The paper introduces the principle of integrated color and spatial relationships feature extraction algorithm and the principle of texture feature co-occurrence matrix, completes the experiment of feature extraction with MATLAB. Then, the paper defines metadata description format of the color texture, spatial relationship features and texture feature, achieves ontology mark for the image low-level features of semantic by MATLAB. Afterwards the paper introduces several typical XML document matching algorithm, establishes a semantic image retrieval system based on ontology model in theory, and test the feasibility of the retrieval system and retrieval performance. Finally, the paper discusses the application fields of semantic image retrieval system based on ontology model.