Zoning documents increases the resolution of indexing from the image level to the field level. A line-delimited tabular document forms a well defined series of regions. However, as image quality decreases, accurate zoning becomes increasingly difficult. Given a sequence of documents with the same layout, we present a robust zoning method which exploits both intra- and inter-document consensus to form a more accurate combined result (template) that can be applied to any other document with the same layout.
Heath E. Nielson, William A. Barrett