Query by fax for content-based image retrieval

Abstract

The problem of query by fax (QBF) and by other low quality images for art image retrieval is introduced. A reliable solution to the problem, the slow QBF algorithm, is presented but it is too slow for fast interactive retrieval situations. It is used as a yardstick for assessing the quality of a much faster technique, the fast QBF algorithm, based on the pyramid-structured wavelet transform (PWT). Both algorithms use the fact that the fax images are almost binary in nature. In the fast QBF method, the query (fax) image is used to identify what would be an appropriate threshold for the database image. This threshold value is then used to select an appropriate set of precomputed wavelet signatures to be used for matching the query and database image. Results of the algorithms are compared. The speed and accuracy of the new technique make it a valuable tool for the query by fax application. © Springer-Verlag Berlin Heidelberg 2002.