Abstract: A new layout-based document image retrieval system is presented in this paper. The system is specifically designed for commercial form retrieval and uses mathematical morphology to extract structural components from the document image. Document layout description is performed by the Radon Transform whereas Dynamic Time Warping is used for matching. The experimental results have been carried out on both real and simulated data sets. They demonstrate the effectiveness of the proposed approach and their robustness with respect to different classes of commercial forms and shifted/rotated document images.
Keywords: Document management, Document Image Retrieval, Mathematic Morphology, Radon Transform, Dynamic Time Warping
