Shape based image retrieval using hybrid legendre moments, moment invariants and fuzzy logic technique

Abstract

Shape is one of the most important features used for Content Based Image Retrieva(CBIR). As compared to its other counterparts such as the color and texture features,shape plays an important part in the recognition of objects by human beings and it is also utilized in Shape Based Image Retrieval (SBIR). The problems in SBIR are normally with its shape based descriptors. Therefore in this study, the problems of the low image retrieval ability in moment invariants and the inability of being rotation invariant for Legendre moments were tackled. Besides that, this study only focuses on the SBIR of binary shape images and the methods employed were region based shape descriptors namely Legendre moments and moment invariants. Moreover, in this thesis, a wide range of comparative study for SBIR using Legendre moments and moment invariants were done. Extensive experiments were carried out by using the 12 randomly selected query images from a database which consist of 1008 shape binary images.