%0 Journal Article
%T Multimodal Markov Random Field for Image Reranking Based on Relevance Feedback
%A Ricardo Omar Ch¨˘vez
%A Hugo Jair Escalante
%A Manuel Montes-y-G¨®mez
%A Luis Enrique Sucar
%J ISRN Machine Vision
%D 2013
%R 10.1155/2013/428746
%X This paper introduces a multimodal approach for reranking of image retrieval results based on relevance feedback. We consider the problem of reordering the ranked list of images returned by an image retrieval system, in such a way that relevant images to a query are moved to the first positions of the list. We propose a Markov random field (MRF) model that aims at classifying the images in the initial retrieval-result list as relevant or irrelevant; the output of the MRF is used to generate a new list of ranked images. The MRF takes into account (1) the rank information provided by the initial retrieval system, (2) similarities among images in the list, and (3) relevance feedback information. Hence, the problem of image reranking is reduced to that of minimizing an energy function that represents a trade-off between image relevance and interimage similarity. The proposed MRF is a multimodal as it can take advantage of both visual and textual information by which images are described with. We report experimental results in the IAPR TC12 collection using visual and textual features to represent images. Experimental results show that our method is able to improve the ranking provided by the base retrieval system. Also, the multimodal MRF outperforms unimodal (i.e., either text-based or image-based) MRFs that we have developed in previous work. Furthermore, the proposed MRF outperforms baseline multimodal methods that combine information from unimodal MRFs. 1. Introduction Images are the main source of information available after text; this fact is due to the availability of inexpensive image registration (e.g., photographic cameras and cell phones) and data storage devices (large volume hard drives), which have given rise to the existence of millions of digital images stored in many databases around the world. However, stored information is useless if we cannot access the specific data we are interested in. Thus, the development of effective methods for the organization and exploration of image collections is a crucial task [1¨C3]. In a standard image retrieval scenario one has available a collection of images and users want to access images stored in that collection, where images can be annotated (i.e., associated to a textual description). Images are represented by features extracted from them. Users formulate queries (which are associated to their information needs) by using either sample images, a textual description, or a combination of both. Queries are represented by features extracted from them and the retrieval process reduces to comparing the
%U http://www.hindawi.com/journals/isrn.machine.vision/2013/428746/