Research On Error Analysis And Calibration Method Of Near Infrared CCD Thermometer

Near infrared CCD thermometer has become a research focus in the field of radiation thermometry,due to its low cost,fast response and the ability of geting temperature field information.The non-uniformity of CCD sensors and optical-system can arouse serious photometric distortion of temperature field measurement.The change of measurement distance can produce great error.Considering the photometric distortion of temperature field measurement aroused by CCD sensors,reasons and characteristics are analyzed.The response non-uniformity of the CCD sensor is mainly caused by the fixed pattern noise,and is characterized by non-uniformity of gain,offset,and response.Based on these characteristics,the segmented non-uniformity correction algorithm is applied in the correction.Experimental results show that gray value of the CCD sensor can fall from 2.37%to 1.15%after correction.Considering the vignetting in optical system,the reason of non-uniformity of optical-system is analyzed.The non-uniformity of optical-system is aroused by vignetting.Optical vignetting is aroused by operture stop on the beam which is not located on the main light-axis.Natural vignetting is aroused by the difference of angle of aperture.The formulas of optical vignetting coefficients and natural vignetting coefficients are established,and verified by the experiments.In order to replace the complex off-line calibration,an estimation algorithm based Particle Swarm Optimization is proposed.Compared with the standard calibration method,the maximum relative error for vignetting estimation by algorithm which based on Particle Swarm Optimization are 1.96%.The impact of change on distance measurement precision is analyzed,the angle of aperature is effected by distance,and the light intensity received by CCD pixel become different,and then the error of temperature is produced.In order to reduce temperature measurement error aroused by measurement distance,one temperature error correction model with distance as independent variable is established.When the calibration distance is 1.9m,and temperature of the black furnace is 1000? and 1100? respectively,the maximum relative error corrected by the model are 0.115%and 0.143%within the distance scope of 1.3-2.8m.And this model can effecitively reduce the influence of distance.The practical effect of the research results on the temperature field measurement of casting billets is analyzed.Experimental results show that gray value of the CCD sensor can be reduced by 1.41%after segmented non-uniformity correction.And the gray value of the CCD sensor can be reduced by 6.40%after non-uniformity of optical-system correction.After tnis two processing,the distortion in temperature field measurement of casting billets is corrected successfully.The temperature compensation of temperature error correction model is-2.1? and it can reduce temperature error in the temperature field measurement.