Research On High Accuracy Temperature Measurement Technology Based On CCD Image Sensing

At present,the traditional contact thermocouple thermometer was used to measure the surface temperature of molten steel in the smelting furnace during the steelmaking process.However,the traditional contact thermocouple thermometer only realized singlepoint temperature measurement.If the temperature measuring instrument was operated in this environment,the measurement results would be lowered.According to the basic theory of infrared radiation,this paper proposed a non-contact temperature measurement technique based on CCD image sensor to measure the surface temperature of molten steel accurately,the solutions to improve the accuracy of CCD image sensing temperature measurement were given.First of all,the non-local mean filter algorithm was used to remove the noise of the molten steel image,the gray value information of the noiseless target image was extracted,for the subsequent establishment of temperature measurement model to provide high-precision data.Then,the least square method and generalized regression neural network to curve fitting the nonlinear mathematic relationship between gray ratio and temperature respectively.However,the temperature must be real-time online display,so commonly used least square method to establish the temperature measurement model temperature measurement to meet the industrial requirements of the temperature interval.Finally,the hardware image selection and software function design of the CCD image sensing high-precision temperature measurement system were studied.The design of on-line temperature measurement display platform was completed by MATLAB software.At the same time,the temperature platform also provided isotherm display,morphological display and pseudo-color images.The traditional measurement results of temperature measurement technology to solve the inaccurate shortcomings,can be further improved after the application.