Implementation of Image Processing in Scanning KTP Data using Optical Character Recognition (OCR)

Abstract

The Indonesian national identity card (Kartu Tanda Penduduk or KTP) serves as the primary identification tool for citizens across various administrative processes, both in the public and private sectors. However, manual data entry from KTPs remains common, leading to potential errors, inefficiencies, and delays. This study aims to develop an Android-based application capable of automatically scanning and extracting data from KTPs using Optical Character Recognition (OCR) enhanced by Convolutional Neural Network (CNN) methods. CNN is applied during the image preprocessing stage to improve text area detection and character segmentation accuracy before the OCR process. The application is developed using the Python, Dart, PHP programming language and designed with a user-friendly interface. Extracted data such as full name, NIK, place and date of birth, and address are stored in a MySQL database via integrated web APIs. The research approach follows a software engineering methodology consisting of requirements analysis, system design, implementation, and system testing. Experimental results show that integrating CNN with OCR increases character recognition accuracy up to 86,7%, especially on low-quality or noisy images. Therefore, this application is expected to serve as an effective solution for accelerating and improving the accuracy of population data digitization in various institutions.