Content Based Image Retrieval System using Boundary Based Shape Descriptors

As the number of digital images around the world gets increased, there is a growing interest in finding images from the large collections. From the large number of images, it is very important for users to retrieve required images via an efficient and effective mechanism. To achieve accurate retrieval results, Content-Based Image Retrieval (CBIR) has become an important research area in recent years. Shape is one of the primary and important low level image features in CBIR used to describe image content. Searching of images using shape features has attracted more attention. In this paper we proposed several boundary based shape features such as area, perimeter, compactness, shape signatures, Fourier descriptors etc. and the study of those features in image retrieval. We have applied the proposed method on a collection of images chosen from MPEG-7 database which is a standard shape database. The image retrieval performance is measured by precision and recall. The proposed Boundary Based shape features and benchmarking database produced the performance metrics of about 70% to 90%.