Tatiana Jaworska

In this article we propose graphical object classification used for image matching in the Content-Based Image Retrieval (CBIR) system containing colour images. The part devoted to image processing and the inner structure of the database are signalled to the extent which is necessary for the reader to understand how the whole system works. Firstly, we discuss the theoretical construction of indexes for the graphical object classification. The indexes are based on feature vectors for each object and spatial relationships among objects for image retrieval. We present some classes and classfied objects. Secondly, we address the problem of the graphical query by example. In order to construct the graphical query we implement the user interface (GUI) which has been developed in the light of humancomputer interaction. GUI enables the user to design their own image which is further treated as a query for the database. The expected reply is a set of similar images presented to the user by the database.