Attribute associated image retrieval and similarity reranking

Existence of countless digital images has given rise to image retrieval in many applications. Conventional image databases being text-annotated pose two major problems of keywords for images and complexity. Hence, retrieval systems based on image´s visual content are more desirable [1]. The content based image retrieval (CBIR) technique, employed here uses visual cues to retrieve images. This technique is query based, extracts the most vital attributes like color, shape and texture. Automatic extraction of spatial based color feature and invariant Fourier descriptors makes it more flexible. The extent of each attribute is obtained from the user, compared with attributes of images in database and most similar images are retrieved based on the degree of similarity.