The Research And Imply Of The Technology Of Content Based Image Retrieval

The popularity of computers and the Internet makes multimedia technology more and more extensive, image retrieval is an important research field of the application of multimedia technology. Using image features of color, texture, shape, and spatial distribution to retrieve image is the main features of content-based image retrieval technology, it gets the results of image retrieval through expressing the lowest lever features of the image’s content and matching them.Color and texture is the most intuitive visual characteristics, they all belong to the low-level features of the image, which is the basis of image retrieval. This paper focuses on the application of these two features in image retrieval. The processing of the color characteristics concludes of the image color space conversion, color quantization and color expression. We do experiments respectively in different images of ten categories using Color histogram in RGB and HSV color space. Through this experiments, we carried out the retrieval results of the same category image in a different color space, and different categories image in the same color space; in texture, according to three texture expression we do experiments in the ten categories of images above with different ways. Through it, we can get the retrieval result of the same category image in different methods and different categories of images in the same way, and then compared them and summary.Using single feature to retrieve image has some limitations. In order to compensate for the inadequacies of it, this paper using the re-integration feature, which concludes color and texture feature in different weights to retrieve image. According to the new method, we do experiment in different categories of images, compared the result and summary.In addition, due to the human’visual is not sensitive to subtle of image information, this paper not only retrieval images from the perspective of image compression and accelerate the retrieval speed through reducing the redundancy of information of image, but also improve the precision of image, whose texture is significant. We design experiments using ten kinds of images of different levels in Pyramid, in horizontal direction, compared the retrieval result of different levels of the same image; in vertical direction, compared the retrieval result of the same level of different images. We can conclude it:for natural images, this method can enhance the image retrieval speed, but for texture image, such as rocks, trees, fabric, not only can enhance the speed but also can increase the precision. So we can use this method to distinguish the various types of rock, or a variety of wood with good effect and good adaptability.This paper does experiment with Mat lab 7.0 and SQL Server 2000, and based on it, provides an experimental platform for research in the environment.