i_OCR is an automated system for processing paper forms of writing of any kind. Part of the image mining suite i_ICONA, it consists of two parts; the general part of the preprocessing and the main part of the recognition.
Specifically, the pretreatment includes correction of the angle of deflection of the document; locate the manuscript page, writing skew correction and segmentation into words. Regarding the recognition we selected recognition by segmentation into characters as opposed to holistic systems that make word recognition without segmentation
Finally, the verification process by crossing with dictionaries or other rules, increases the accuracy of the system during the export of data so the user can easily confirm it by pressing a button. The dictionaries are created by DIRECTING for each case separately and are part of the application