Image Retrieval Technology In The Cg Project Management System

With the rapid development of network technology, multimedia technology, database technology and computer vision technology, the image, as an important multimedia information carrier, has been widely used in many fields. In order to facilitate users to find their own image quickly and accurately from the vast resources of the images, content-based image retrieval (CBIR) technology is gradually becoming a hot research.Based on the research and master of the CBIR technology and the knowledge in related fields, this paper has studied the extraction method of the image's characteristics, and completed the following work:1. This paper uses the HSV color space that close to human perception to pick up the image characteristic. After studying several characteristic extraction methods, we have proved the effectiveness of the above approaches through experiments, compared and analyzed their performance as well. And finally we adopted the updated local accumulative histogram. Based on region-based and color-based researching of the shape, this paper has improved the traditional similarity measurement, realized a new method of back-shaped block. Experiments showed that this method can increase accuracy.2. To improve performance of the retrieval system, we have studied three kinds of image retrieval methods, color and texture characteristics, texture and shape features, color, texture and shape features, and realized a comprehensive image retrieval system using local accumulative histogram method, gray co-occurrence matrix and back-shaped block and several algorithms, Through experiments we can show that the retrieval methods could effectively improve the accuracy of the system.3. Feed back system is imported in the comprehensive image retrieval system, the main idea is adjusting the weight of image characteristic dynamically to change the result of retrieval through the feed back information by consumers, and then the image is searched by new weight. The experiment shows that the correlative feed back strategy can enhance efficiency of retrieval in the system, and can search image for consumer more efficiently.4. Finally, the image retrieval technology is used in the resources searching module of CGProject. ImgSearch, an image retrieval system, is designed and fulfilled. It optimizes the function of the image retrieval, and experiment show that the system has the perfect practicability in the CGProject.