Title :
Robust recognition of documents
Author_Institution :
Siemens AG, Munchen, Germany
Abstract :
To make interpretation of scanned documents much more robust, the system IDA uses different knowledge about geometry and text. This knowledge is handled by an analysis process, which is relatively free in what to do and how to do it. These degrees of freedom are filled by utilization of control knowledge. Decisions are made with regards to context. The analysis process calculates truth values as well as numerically based test candidates. All analysis states are stored in justification networks similar to truth maintenance systems. This is used to support the solution of inconsistencies and ambiguities. The results of analysis as senders, dates, and authors are reliable enough so that they can be used by storing documents in databases
Keywords :
document image processing; image recognition; knowledge based systems; optical character recognition; ambiguities; control knowledge; inconsistencies; justification networks; scanned documents; truth maintenance systems; truth values; Cities and towns; Geometry; Image databases; Image recognition; Maintenance; Research and development; Robustness; Shape; Text recognition; Tin;
Conference_Titel :
Document Analysis and Recognition, 1993., Proceedings of the Second International Conference on
Conference_Location :
Tsukuba Science City
Print_ISBN :
0-8186-4960-7
DOI :
10.1109/ICDAR.1993.395698
