World Wide Web CBIR Searching Using Query by Approximate Shapes

Nowadays more and more images are stored in the World Wide Web. There are a lot of photo galleries, media portals and social media portals where users add their own content, but also they would like to find the proper ones. The problem of searching for an image is not trivial. Objects present on images may have e.g. different colors, backgrounds or orientations. Moreover, the image may contain many other details which may be hard to be described by words. This paper presents a new system which may be used to query for images from the internet which is based on our Query by Approximate Shapes algorithm. The main idea of the proposed approach is to gather images from the internet. Next, all images are processed using our algorithm which is based on decomposing objects into a set of simple shapes. During the query, depending on its type, an example image or a sketch is used. For both types a graph is constructed which is compared with graphs in the database.