Document layout extraction using soft ordering
Phillip E. Mitchell, Hong Yan · Optical Engineering · 2002
We present an algorithm that can determine the layout of an arbitrary document with great flexibility. The bottom-up approach of pat- tern extraction and classification provides good segmentation and is in- sensitive to skew. Soft ordering is a feature that improves segmentation by allowing distinct regions to physically overlap. It is also used to deter- mine the correct order of the document regions. The algorithm can ex- tract and place all the distinct document regions into a logical layout and column structure. © 2002 Society of Photo-Optical Instrumentation Engineers.