Abstract: A template image of a form document is processed to identify alignment regions used to match a query image to the template image and processing regions to identify areas from which information is to be extracted. Processing of the template image includes the identification of a set of meaningful template vectors. A query image is processed to determine meaningful query vectors. The meaningful template vectors are compared with the meaningful query vectors to determine whether the format of the template and query images match. Upon achievement of a match, a homography between the images is determined. In the event a homography threshold has been met, information is extracted from the query image.