ABSTRACT

Content Based Image Retrieval, in the current scenario has not been analyzed adequate in the existing system. Here, we implement a prototype system for web based image retrieval. The system is based on description of images by lexical chains which are extracted from text related images in a web page. In this paper, we provide Relevance Feedback (RF) techniques that aim to the real world user requirements. The relevance feedback techniques, based on image text description are expanded to support image retrieval by combining textual and visual features. All the feedback techniques are implemented and compared with precision and recall criteria. The experimental results prove that retrieval methods that makes use of both text and visual features achieve overall better results than methods based only on image’s text description.