In this paper, firstly, two image index methods based on weight and center of image gravity are described briefly, so that a method noted IGDC is given. After that an efficient image index technique is provided through R-tree. Secondly, two kinds of distance about characteristic vector and the structure about the rectilinear rectangle that is called the minimum bounding rectangle (abbreviated to MBR) are defined, then a new image index method based on idea of R-tree is designed further by utilizing the IGDC structure. The experimental results show that this method is very efficient, at the same time, the method is a attempt about defining multi-distance in image sets. Thirdly, a calculation method about the interest point of image is described through the feature about singular value of image and a signaler value index method is introduced. The method is not sensitive to images with noise.In the end , a fractal index method is discussed. The search algorithm about FAI sequence and the transformation between FAI sequence and Hilbert sequence, a kind of important image index method, are proposed . It will be developed greatly in the future. |