Improvement the character recognition of postal address with matching method

OData support
Dr. Szűcs Gábor
Department of Telecommunications and Media Informatics

Nowadays thousands of letter mails are sent around the world and the number of them increases day by day. The manual processing of these items takes many times. To solve this issue, namely to recognize the delivery addresses of these letter mails OCR (Optical Character Recognition) engines have been implemented into computer systems. These systems are capable of extracting the texts out of images, but they make mistakes like recognition of unnecessary signs and characters, and they recognize some characters inappropriately. Our implemented system automatically recognize the name of the settlement and its zip code from inaccurate OCR engine’s output, and to improve the recognition with text matching method (e.g.: Levenshtein distance).

Letter mails are tracked by barcodes during delivery. The computer system has to be able to identify the mails and send them to the right direction, to the correct address on the ribbon when they arrive to a decision point. This thesis discusses solutions for barcode detection, which scans the string (character sequence) of the barcode, and decodes the barcode optically by a barcode reader, after that matching is developed between barcode and this string.


Please sign in to download the files of this thesis.