Templategeneration from Postmarks Using Cascaded Unsupervised Learning

A method for automatically extracting templates for each category of these postmark stamps is described. The problem is complicated by the high levels of degradation present in the cards. The ink is faded, and the paper is yellowed. A significant quantity of both occlusion and dropout exists in the same images. The postmark stamps overlap the other content on the card. The rubber stamp sometimes did not make full contact with the paper resulting in an incomplete image. Excesses of ink connect the text and line components. The approach uses a cascade of unsupervised learning steps separated with image cleaning. The templates once extracted can be used to group the postmarks, and will contribute information about the postmark content to better separate from the paper and other interfering marks to extract further information about the postmarks and postcards.