Bill Denomination Recognition Using Digital Image Analysis

摘要

This research talks about a system that can recognize denomination of bills particularly in the country of the Philippines using digital image analysis. The system can be used in Cash Accept Machines in which people who are cash depositors interact with machine directly. The initial algorithm of the system has some steps such as; grayscaling, binarization using Otsu method, region of interest, and then optical character recognition. Based on the results, the accuracy of the system is 96.67% for denomination recognition and 98.59% for digit recognition. To reach to the highest accuracy of 100%, based on experimental results and the digits used in Philippines’ bills, the researchers add a conditional step to the algorithm in which the digits change to other digits. From software view, the researchers used Linux Ubuntu 18 as operating system, Python as programming language and OpenCV as library. From hardware view, the system was implemented in a computer having CPU Core i5, 4GB RAM and 1GB VGA.