Matching images with textual document using TFIDF method

Image is used in any field, starting from newspapers, magazines, web sites, movie posters, up to scientific researches to enhance the story effects of the images. Editors, journalists, brochure illustrators, taxonomists in biological science, DNA research scientists, star classifying astronomers rely heavily on images appropriate with the accompanying texts. Sometimes the images themselves are more important than the accompanying texts. Then it is very important to find the best images to illustrate the text. This research consists of two stages, namely database forming stage and text-based image retrieval stage. As a tool to build a database for keywords or texts or documents annotated images, MIRA (Multimedia Information Retrieval Application) is employed. An application of TFIDF algorithm into MIRA is used to determine the weight of a term or word in a text or document that annotates an image and then compared to keywords extracted from a text or document. In this study, we compared the results of image search using Google search engine with our system, MIRA. Based on the results of text-based image retrieval using keyterm, it can be concluded that image search is successfully distributed. The developed system which is text-based image retrieval is capable of image searching which suit the hopes with optimal results.