The Design Of Liquid-in-glass Thermometer Verification Device Based On Image Recognition And Wireless Monitoring

Liquid-in-glass thermometer has a huge amount usage and it is the most commonly usedThermometer in the production and living. According to The Metrological Law of thePeople’s Republic of China, the liquid-in-glass thermometer is subjected to compulsoryverification in some cases. In the heat engineering departments of our test unit, dailyverification work the liquid-in-glass thermometers accounting for the vast majority of theworkload of verifier, because the verification process need to take almost40minutes to anhour. The verifier, must to stande beside the verification device, reads and records the relevantdata when the thermostatic bath reaches a specified temperature, then finishes the test report.It is time-consuming and energy-consuming, low-efficient.Based on image processing, computer vision, pattern recognition and wireless networktransmission research, this paper proposes and designs a fully automatic liquid-in-glassthermometer verification device, provides the solution of its real-time video monitoring.Expect to use the computer to replace the manual to complete the verification ofliquid-in-glass thermometer.This paper describes the software design and implementation of the verification systemand real-time video monitoring system of this device.According to the basic idea of themodular design, the software system described in this paper is divided into the followingmodules：The camera control and image capture module based on the DirectShow forwindows.Image processing module that developed under the OpenCV computer visionlibrary. Character recognition module used the shape-contextalgorithm and the thin-platespline transform. The wireless real-time video monitoring moduleused the X264and live555open source library.Because the wearafter long-term use，uneven illuminationof the image acquisition devicearound,the characters in the image of the liquid-in-glass thermometer that camera capturedmay be incomplete. These make the correct rate of character recognition reduce, andthese arethe key problem that restrictingthe developmentof some fullyautomatic verification device. Toovercome this problem and improve the correct rate of the incomplete character recognition,we take the shape-context algorithm as the core algorithm to recognize the incompletecharacter in this paper, and then achieved good recognition effect. The experiments also showthat the real-time video monitoring system described in this paper has a good performanceand fluencyon the premise that ensure real-time and definition, It can meet the monitoringrequirements of verifier in the office floors with complex terrain.