| With the development of the multimedia network technology, the application of the image is extensive, the rapid increase of image application, Content-based Image Retrieval (CBIR) becomes most active one in multimedia retrieval field. In order to analysis the information included in an image, the CBIR system always analyses the color, texture, shape, and other low-layer image features, to establish retrieval vectors as retrieval index. In present time, the main CBIR method is similarity retrieval based on multi-dimension feature vector of image. Extracting features from image is the key issues in CBIR.In this paper, on the base of widely referring to the material about Content Based Image Retrieval of home and abroad, the present and feature of CBIR and the key issues are discussed, and some methods of extracting image content are analyzed in detail. This dissertation deals with the extracting features from image theory and its processing techniques, some methods for extracting features from image based on vision content are discussed. In this dissertation an extracting features from image based on content is designed, the experiments show that the color, texture, and shape features proposed in this dissertation are effective in the description of image content. The color feature, being important vision information of image, is often used to describe image content and has been used broadly. In this dissertation we have discussed key questions on how to use of the characteristic of color, which include expressing color, obtaining the characteristic of color. The method of extracting color feature based on image content is analyzed, color histogram built from cumulative distributions of content colors is a main color feature using in image retrieval. Based on extracting color feature from BMP static image, color histogram is adopted. To show color characteristic of image, the method of the HSV color space, which is suitable to the visual characteristic of human, is utilized. Taking advance of human's feeling to color, it quantifies the color set with unequal interval, and get characteristic vector, regarded as color feature of image. It is natural that image under observation is divided into the object and background by human vision. The image retrieval performance evaluated mainly with the object-based measurement. In this dissertation, considering the image characteristics, on the basis of mathematics substation thinking, we extract color feature for each block and a weight color histogram method are presented, which can emphasize the color feature of the object in the image, it can make the color feature to have the spatial comparability. By comparing with general color method, the color image retrieval result accords with the better human vision perception. The improvement of retrieval efficiency is proved by experiments.However, the color histogram does not contain the information of spatial distribution of color across an image. So there might be different images, which possess different contents but share with same color histogram. Because the ability of expressing image semantic information with the color feature is too limited, so a texture feature is presented. Texture is an important image feature that is hard to describe and has no acknowledged precise definition yet. Texture, as an important component of the human vision, can reflect the depth and surface information, and supply the human vision with the recognition and understanding information.In this dissertation, we discuss the method of extracting texture feature from image, two kinds of method of extracting texture feature are adopted, which are based on co-occurrence matrix and Gabor filters respectively, and the image retrieval results based on the two feature vectors are compared later.First based on co-occurrence matrix method, we get energy, moment of inertia, entropy, correlation, local calm five statistical values from the four directions(0 degree, 45 degree, 90 degree, 135 degree) co-occurrence matrixs of the...
|
PDF Full Text Request