Curvature scale space image in shape similarity retrieval

In many applications, the user of an image database system points to an ima ge, and wishes to retrieve similar images from the database. Computer visio n researchers aim to capture image information in feature vectors which des cribe shape, texture and color properties of the image. These vectors are i ndexed or compared to one another during query processing to find images fr om the database. This paper is concerned with the problem of shape similari ty retrieval in image databases. Curvature scale space (CSS) image represen tation along with a small number of global parameters are used for this pur pose. The CSS image consists of several arch-shape contours representing th e inflection points of the shape as it is smoothed. The maxima of these con tours are used to represent a shape. The method is then tested on a databas e of 1100 images of marine creatures. A classified subset of this database is used to evaluate the method and compare it with other methods. The resul ts show the promising performance of the method and its superiority over Fo urier descriptors and moment invariants.