FORM PROCESSING AUTOMATION FRAMEWORK USING ZONAL OPTICAL CHARACTER RECOGNITION AND WORKFLOW METHODOLOGY

Abstract

Automated form processing has yet to become mainstream in the industry. In the small to medium business landscape, such process is unlikely to be used due to the high cost of initial investment. Some of the readily available solution require a high upfront payment or a recurring subscription. However, despite the availability of fillable forms, many businesses still rely on printed and scanned documents which create problems for any automated OCR solution. For instance, a scanned document may have different orientation or position from the original document which makes the process more prone to error. A preprocessing step must be taken prior to process these documents to correct for these factors. However, not all businesses have the needed skilled mind or the time to design an in-house optical character recognition (OCR) process. And because of these technical challenge, these business ended up resorting to manual labour which will cost more in the long run compared to an automated process. Therefore, much more affordable OCR solution which can be easily adapted to fit certain business needs is sorely needed. Accordingly, this thesis aims to rectify this by providing a customisable zonal OCR framework with preprocessing capabilities. It will perform bitmap processing and parsing on an input bitmap obtained from a given document. Additionally, a workflow-oriented approach will be employed to provide sequential data parsing and processing logics. This will allow processes to take input from the initial state or previous process and return an altered bitmap for further processes where applicable.