Research On Content-based Image Retrieval Technology

With the rapid development of network technology, the rapid growth of imageinformation, digital images from the mass rapid extraction of useful information hasbecome an urgent need for life and work. Content-based image retrieval technologyin the1990s, emerged as the field of intelligent information processing researchhotspot.Traditional image retrieval process relying on manual text annotation on theimage and then keyword to retrieve images, this method is time-consuming, but alsothe lack of reliability, initiative too. Content-based image retrieval to overcome theshortcomings of traditional search, image search directly from visual features to bestarting in the image database to find similar images with them.Content-based image retrieval through the image visual features (such as color,texture or shape) to represent the content of the image, and the image obtained fromvisual features as image content index. Content-based image analysis does not aim toreally understand the semantics of the image, but the image analysis demerit extractsome with strong partitioning capabilities and as close to the image semanticcharacteristics, the use of these features as a matter of image similarity index ofretrieval.This paper discusses the results of domestic and foreign research status andcontent-based image retrieval technology to be introduced content-based imageretrieval system, analyzes and studies the image retrieval key technologies, includingfeature extraction, similarity measurement and performance evaluation standards.Which feature extraction techniques and similarity measure techniques are based onthe core issues are content-based image retrieval technology. In this paper, these twoissues, how to describe the image content, accurate extraction of features, as well asthe precise content of the image similarity measure.This paper focuses on the retrieval method based on color and shape featuresproposed block color histogram retrieval method and rely on the feature points todetermine the outline of the image retrieval method based experiments were verified by two features based on color and shape feasibility feature retrieval methods, andfinally on the basis of the above two methods, we propose a method for extractingimage edge information using color features. Under normal circumstances, there aredifferences between the colors on the color image of the object and the background isinevitable, use these color differences on the shape of the edge information to extractobjects more easily than direct extraction of edge features. This paper studies thedifferences by changing the color of the image to extract the edge information of theimage objects for shape-based image retrieval. Finally, the experimental verificationof a single color feature and shape features rely more effective than the previouslyused this approach and practical.