Graphic Form Recognition - Read-Out Forms Reliably

As part of the classification of text, the reading of forms through automated processing presents a special discipline. Poorly scanned documents or fax entries with low image resolutions are rarely processed by the traditional OCR system. To overcome these problems, the iForm solution from ITyX abandons form reading through OCR, using instead the cognitive process of AI.

Complementing the iForm processes are manual definitions of text areas for the zonal reading of data. In addition, it returns askew, compressed or rotated documents back to the original document layout in the system. A maximum reading quota is achieved through the combination of the processes.

In mechanical form recognition and clasification of running text it is important to keep the 'False-Positive-Rate' as low as possible - ie the porportion of false classifications in automatically categorised processes. On request, unstable results are automatically moved to manual human post processing with Mediatrix MAILROOM. Defining the threshold enables very fine control. In the classification matrix, an optimum tolerance of errors with regards to process quality can be set for each business process. In this way it is, for example, possible to reduce temporarily the error tolerance in aid of a higher throughput in overload situations. 

ITyX Technology for Reading Forms includes eg:

• Rapid implementation of form-bound processes through visual recognition (training process)
• Definition of zonal scope with decreasing tolerance
• Optimal false-positive rate