Desenvolvimento de uma biblioteca para descrição de imagens

Abstract

With the popularity of digital photography and the World Wide Web, it was increased the demand for image retrieval systems, especially the Content-based Image Retrieval (CBIR). The effectiveness of a CBIR system is influenced by the choice of descriptors, distance functions and similarity operators used, thereby making it extremely important performance testing to make the best possible combination for each case. This final graduation thesis aims to describe the process of developing a library for describing images that allow the easy attachment of new descriptors, distance functions, similarity of operators and performance tests; and a user interface which allows the interaction between the end user and the library.