Retrieval by visual content in image databases

Large amounts of pictures and videos are created, published, transmitted and accessed everyday by corporations and the general public for different uses, from entertainment to mobile robotics. Multimedia content has become a vital enterprise asset and visual content management has emerged as a strategic necessity. Content-based indexing of visual databases is a key technology for representing huge databases of images. A novel image retrieval technique, based on the Polar Weighted Walktrough (PWW) representation, is presented and encouraging results of searching by content in image databases are shown. The proposed metric, has very important properties: it is reflexive, invariant with respect to image scaling, compositional and, in particular, invariable to rotations.