Automatic detection of change in address blocks for reply forms processing
Keerambur Ramaswami Karthick, Stephen Marshall, Alison J. Gray · Strathprints: The University of Strathclyde institutional repository (University of Strathclyde) · 2008
Abstract — In this paper, an automatic method to detect the presence of on-line erasures/scribbles/corrections/over-writing in the address block of various types of subscription and utility payment forms is presented. The proposed approach employs bottom-up segmentation of the address block. Heuristic rules based on structural features are used to automate the detection process. The algorithm is applied on a large dataset of 5,780 real world document forms of 200 dots per inch resolution. The proposed algorithm performs well, with a detection accuracy of 98.96 % and an average processing time of 108 milliseconds per document. Index Terms — Address correction detection, Document form processing, Document image analysis, Off-line scribble detection I.