Content-based Image Retrieval Methods And Key Technologies

With the rapid development of multimedia technology, its main content as images or video is also increasing. In particular, image data on the Internet becomes more and more, content-based image retrieval (CBIR) has appeared in order to effectively and quickly query and retrieve images in image database. The method automatically extracts its feature vectors for a given sample image in image retrieval, and makes its features comparison with the feature vectors of the database in accordance with the similarity measure, and then retrieves the most similar images with the sample image from the image database. Thus, the advantages and disadvantages of the feature extraction for an example image directly influence the retrieval results.Content-Based Image Retrieval technology has been ten years of research history, and achieved many research results, but still unable to fully meet the customer's requirements. One of the most important problems is the gap between the low-level visual features of image content and high-level semantic when the user retrieves, which is a difficult problem in computer vision, image understanding and pattern recognition and so on. The study of the low-level visual features of image content, as well as the image similarity measure are meaningful, and its application areas is wide but also faces challenges.In this paper, content-based image retrieval techniques have been studied, mainly include the followings:Firstly, it elaborates the theories and developments of the content-based image retrieval, and discusses in detail some key technologies, indexing and evaluation criteria of the retrieval results.Secondly, using color features and texture features of image retrieval methods of the content-based image retrieval has been made the systematic analysis, and focuses on the characteristics of several colors and the extraction technology in different color space, and based on the Gray Level Co-occurrence Matrix Texture feature extraction techniques. Then the algorithm of some features of color and the gray level co-occurrence matrix has been implemented. Because of the characteristics of a single color or texture feature extraction algorithm has its own limitations, and therefore the retrieval results should not be the largest user needs.Finally, the paper proposes the image retrieval algorithm based on the fusion characteristics, which combines the color features with the texture features in order to retrieve images. Because the algorithm not only consider the plentiful information of image color but also the texture structure information of the image in detail, the combination of both makes the retrieval results better ,and precision and recall are higher.Content-based image retrieval involves a wider knowledge, this paper focuses on two key technologies and their fusion of the two carried out a detailed discussion and implementation, and will continue deeply to research and explore in more aspects of the characteristics.