Technological and dynamic growths of digital media have increased volume of the multimedia corpus. Due to this growth, it is strongly urged for a system that can efficiently and effectively compiles the user demand, and retrieving the relevant images. Keyword based system retrieves an image on syntactic matching, i.e. string matching not concept. Content Based Image Retrieval (CBIR) systems retrieve the image based on low level features and still exist a gap is called semantic. This paper discussed snippet technique that covers the semantic gap as well as Word Sense disambiguation problems. It extracts user queries for expansion with the help of Knowledgebase WordNet and ConceptNet. Experiments performed on the open benchmark image dataset LabelMe. A substantial improvement has been achieved in terms of precision and recall. Remarkably outperformed of Results and showed 84 corrects. © 2005 - 2012 JATIT & LLS. All rights reserved.