OPTICAL CHARACTER RECOGNITION IMPLEMENTATION FOR ADMISSION SYSTEM IN UNIVERSITAS PERTAMINA

Abstract

Starting in 2019, prospective college students require to take Computer-Based Writing Exam (UTBK) to register for the state universities in Indonesia. Some private university also adopts this exam as a requirement for admission. One of the private university that adopts it is Universitas Pertamina. UTBK consist of several exam group score printed in a digital certificate in image format (jpg). The university admission team must download the UTBK certificate that has uploaded by applicants, read and record the score for each exam group then make a calculation to make a decision whether the applicant is accepted in a certain school in the university. This research proposes to replace the manual process performed by the admission team with optical character recognition (OCR). The OCR engine will extract text from an image. Some information from the extracted text is calculated to provide an acceptance decision. The research shows that OCR cannot accurately convert text from an image when there is a grayscale background in the image. However, image preprocessing can improve overall accuracy. Lastly, Tesseract performs better in converting black text with white-background than white text with a black background.

Keywords

• artificial intelligence
• computer vision
• pattern recognition
• machine learning
• optical character recognition