Content-based Multimedia Information Retrieval Research

With advances in the computer technologies and the advent of the World-Wide Web, there has been an explosion in the amount and complexity of various multimedia data being generated, stored, transmitted, analyzed, and accessed. In the face of such voluminous and complicated information oceans, it is pressed for that one can fast and efficiently acquire the needed information, and powerful multimedia information retrieval tools and presentation tools are absolutely necessary Content-based image retrieval and multimedia synchronization models are currently two active research fields, and the researches have very important theoretical and practical meanings.Image segmentation is the foundation of image feature extraction. In this dissertation, We describe the texture spectrum features of image, and propose a kind of image segmentation method based texture spectrum features. It combines texture spectrum features with region-growing algorithm and implement unsupervised texture segmentation. The texture spectrum features have the advantages such as the sensitivity to direction, and experiment shows this method achieves good segmentation results.We describe the implementation principles of texture-based image retrieval, which includes co-occurrence matrix, image texture feature extraction from co-occurrence matrix, rotation-invariant feature extraction and the retrieval system design.Due to the gap between high level concepts and low-level features and the subjectivity of human perception, the traditional computer centric approach has severe limitations. To overcome this kind of limitations, we propose the interactive image retrieval algorithm using relevance feedback technique. We also describe the image object model, feature normalization...