Automated form processing system based on Sobel operator

摘要

Automated form processing is a procedure that can dramatically accelerate system performance, reduce the proportion of human work and interventions, and is applicable in many areas. The aim of this paper is to describe the procedure that successfully locates various fields for data entry on scanned documents and prepare their content for further processing — segmentation and recognition. The procedure starts with preprocessing to make data entry fields easier to identify. It consists of contrast adjustments and corrections to the text skew that can occur when scanning. Data entry fields are then identified by a developed algorithm using the Sobel operator. Finally, field contents are extracted from the original image. The procedure performance in data extraction from the form is tested on a number of scanned and computer-generated forms in A4 format, giving good accuracy. This procedure gives satisfactory results for many applications of automated form processing.