A Preprocessing Algorithm to Increase OCR Performance on Application Processor-Centric FPGA Architectures

摘要

The aim of this research is to build up a fully automatic preprocessing algorithm capable of binarize and dewarping digitized documents, embedded in an application processor-centric Field Programmable Gate Array (FPGA), in order to develop an autonomous voice scanner for blind and visually impaired. Providing for blind the ability of hearing books without further assistance is the main purpose of this work overall. This is a part of a larger project, called "The Vocalizer Project", emerged due to a demand by Brazil's Ministry of Culture and Education for utilization in schools and public libraries, and is addressed for having more inclusive and intelligent cities. Furthermore, it is destined for the inclusion of blind and visually impaired people to the vast bibliographic material existent.