Application of Internet of Things module and optical character recognition algorithm to automatically extracting readouts of radiation survey meters

Abstract

With the development of Internet of Things, radiation measuring devices today are embedded with wireless network connections. However, many devices being used in facilities lack this feature, making it difficult to remotely access data as well as to integrate into a multi-point radioactive monitoring system. In this study, we designed a system called RadOCR that extracts the readouts displayed on the LCD screen of Radiation Alert Ranger® survey meter in real time. The system consisted of a device attached to the meter and a software deployed on a computer. The RadOCR device consisted of a ESP32-CAM module which took photo of the meter’s LCD and sent to the software via a WiFi connection on a local network. The software utilized image processing algorithms from OpenCV and the optical character recognition Tesseract OCR model to extract the readouts, then showed them on a graphical interface, and saved them with timestamps to a csv file. The test results show that the system operates stably with an accuracy of (97.365 ± 0.082) %.