A hybrid approach to detection and recognition of dashboard information in real-time
Yu Tao, Yong Yue, Paul Craig · 2017
The digital industrial dashboards are developing rapidly. But only to display numbers in dashboards has not met industrial demands, more text information are added into them. Therefore, this also increases the workload for the instrument testers. Traditionally, for testing the stability and accuracy of the displays, most testing work relies on human observation. In addition, some errors would be caused by fatigue, carelessness and other uncertain factors during human observation. In order to improve the efficiency of instruments reading and recording, this paper proposes an end-to-end real-time instruments information location and recognition method based on OpenCV and a popular optical character recognition (OCR) engine, called “tesseract”. We also design an online testing strategy to verify what is the performance about this system. Finally, the result would be generated as a test report for the display tester.