一種軟基于彩色CCD圖像傳感器的溫度場(chǎng)測(cè)量系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)

1、Design and Implementation of a Soft-measuring System for High-temperature FieldBased on Color CCD Image SensorLI Yuxiao, LI Sheng, and Ye KuntaoDepartment of Science, Jiangxi University of Science and Technology, Ganzhou 341000,ChinaThe soft-measuring system consists of an optical lens, a color CCDc

2、amera, an image acquisition card, a computer and associatedmeasuring software based on colorimetric temperaturemeasurement. Using the system, the temperature distribution of ahigh-temperature field can be measured and then displayed on thescreen. A method of adjusting the intensity of light is devel

3、oped toexpand the range of the measuring temperature of the system, andthe calibration experiments through blackbody furnace are adoptedto calibrate the soft-measuring system. The errors of themeasurement caused by the non-ideal spectrum response of CCDimage sensor are corrected. M ulti-spectral seg

4、menting is applied onthe image of the high-temperature field, thus various high-temperature noises in the color image can be reduced or eveneliminated. The experiment indicates the soft-measuringinstrumentation meets the high-temperature detecting requirementsof industrial production due to its high

5、 accuracy and reliability. IntroductionIn the process of melting, melt temperature is a very important parameter which directly affects the quality of melting products. So scientists in all countries have concerned the measuring temperature technology in modern metallurgy process, and did a lot of r

6、esearch work, obtained some achievements 1-2. The United States FLUKE company is a professional electronic testing tools manufacturer, its representative products in measuring temperature field are infrared thermography and infrared thermometer. Thermography is mainly used in equipment fault detecti

7、on and maintenance, it always works below 600C; the highest measuring temperature of infrared thermometer may be up to 1000C, but everytime it can only measure temperature to one point, not to the whole temperature field.Compared with the traditional measuring temperature technology ,radiation measu

8、ring temperature has many advantages such as the range of measuring temperature is theoretically unlimited, measuring temperature instruments dont directly contact with the tested object so its temperature will not change, the response time is short, the temperature field distribution of objects sur

9、face can be measured ect, therefore, the soft-measuring system for high-temperature field based on color CCD image sensor becomes a research hotspot in high-temperature measurement field recently 3-4. Our study shows a high-temperature field measurement system composed with an optical lens, a color

10、CCD 10.1149/04533.0027ecst The Electrochemical SocietyECS Transactions, 45 (33 27-36 (2013camera, an image acquisition card, a computer and associated measuring software. This system has high accuracy and reliability, also meets requirements of high temperature detection in industrial production. Th

11、e Principle of Colorimetric Temperature MeasurementWhen the temperature is lower than 3 000K and the wavelength is less than 0.8m, Planck formula can be approximately placed by Wien formula, so monochromatic radiation brightness function of the radiator can be represented as follow:(T C e C T L 251,

12、= 1Here, C 1 and C 2 are the first and second radiation constant respectively; L (, T is radiator monochromatic radiation brightness; T is the actual temperature of the radiator. The channel output values of three primary colors in color CCD and radiator monochromatic radiation brightness keep a lin

13、ear relationship within a certain range.Suppose CCD works in linear area, its spectrum response characteristics are idealized to monochromatic response, the R, G, B three-channel output values can be get by mean value theorem of integrals:(=T L Ab d r T L A B T L Ag d r T L A G T L Ar d r T L A R b

14、b g g r r ,104,104,104,7108.7108.37108.7108.37108.7108.3777777 2Set individual K parameter for R, G, B, we have:(=T L K T L Ab B T L K T L Ag G T L K T L Ar R b b b b g g g g r r r r , 3Here, A is CCD response characteristic parameters, r (r 、g (g 、b (b are respectively CCD spectrum response charact

15、eristic function to R, G, B three primary colors signal. Color CCD wavelength response range is located in visible light area from 390nm to 780nm. Take r =700. 0nm,g =546. 1nm,b =435. 8nm, colorimetric temperature measurement equation can be computed by using the expression of R 、G in equation 3 5:(

16、T T K G R C T g r r g r g ,ln ln 5ln 112+= 4Here, is CCD response coefficient. Most of actual object spectrum emissivity dont change big in a narrow band range, so it can be approximately thinked that . By calibrating experimental data the fitting curve relationship between ln (R /G and K value can

17、be determined. In actual temperature measurement process, the R, G value of every pixel can be get from radiation image, and the corresponding K value can be calculated according to the fitting curve, then temperature field distribution can be obtained by equation 4, at last temperature measurement

18、results are corrected by smoke error compensation and displayed by false color temperature field.It is worthy pointed out that the spectrum response range of color CCD camera is from 380nm to 780nm in visible light area and only when the temperature is higher than 1073K visible light band will have

19、more radiant energy 6, therefore, temperature measurement instruments based on CCD are mainly used in high-temperature measurement with higher than 1073K temperature in order to improve its sensitivity. The Hardware Structure of SystemColorimetric temperature measurement system based on color CCD is

20、 mainly composed with imaging system, image sensor system, image processing system (image acquisition card and computer, as shown in Figure 1. Figure 1. The Diagram of Temperature Measurement System Based on Color CCD(1 The camera lens gathers light signal from radiator surface to CCD photographic t

21、arget surface. The focal length and aperture of lens can be adjusted.(2 Color CCD camera can complete photoelectric conversion process, output three primary colors signal included brightness information and convert them to standard video signal. Now CCD image sensor is mostly used as a camera equipm

22、ent, so it usually has functions of electronic shutter, AGC gain and correction, which are very different from functions temperature measurement needed. Electronic shutter mode should select the fine-tuning 1 789 level in order to get moderate exposure. Automatic gain control (AGC in temperature mea

23、surement process should be closed in order to eliminate signal distortion caused by AGC. correction circuit in CCD camera causes that the relationship between output current I and input light intensity E is nonlinear index as I = ( E t , but this is not allowed in temperature measurement process, so

24、 the value must be set to 1.0. (ln r g K K K =ln (, (, 0r g T T =Automatic white balance circuit can destroy the corresponding relationship between R, G,B signal and temperature, so it should be closed.(3 Image acquisition card real-timely collects standard video signal from color CCD and implements

25、 A / D conversion in order to convert analog video signal to digital image signal, this facilitates computer to store and process signal.(4 The computer stores digital image signal, completes measurement temperature calculation and temperature field processing, also extracts relevant temperature inf

26、ormation from the calculated temperature field.This system uses Japanese COMPUTAR companys M2514-MP optical lens of which aperture and focal length can be manually adjusted; CCD image sensor is Japanese JAI companys CV-M77 color CCD camera with line-by-line scanning, which has variable electronic sh

27、utter, controls CCD shutter time by programming to change CCD input light, and finally automates temperature measurement process; image acquisition card used is Canada CORECO companys high-performance PC2-Vision with PCI interface.PC2-Vision image acquisition card includes built-in 8MB memory used f

28、or buffering image data between camera and host system. On-board memory can ensure that image information will not be lost because of PCI bus delay during transfering to system memory. The image is firstly captured in on-board memory, then quickly moved to host system for processing or displaying, i

29、ts transmission rate can up to 100MB/s, so memory linear addressing can realize maximum utilization. The Software Structure of SystemImage processing software is programmed by Matlab, and it shoule be used on the basis of accurate calibration, so it is designed to two work modes: semi-automatic cali

30、bration and automatic temperature measurement. Semi-automatic calibration needs artificial participation to input actual temperature of blackbody furnace and calculate the fitting curve coefficient by numerical approximation algorithm and so on. The work flow is shown in Figure 2. In temperature mea

31、surement process, image is taken by CCD and filtering, segmentation and morphology processing based on high-temperature radiator color information are implemented on it, then R, G value of every pixel in image are used by soft measurement model of high-temperature field to calculate its surface temp

32、erature field distribution and judge whether the radiation image is interferenced by smoke, with regard to smoke image, it must be processed to remove interference, finally the surface temperature field distribution of high-temperature radiator shoule be displayed in false color way. Figure 2.The Wo

33、rk Flow ChartA. Automatic Adjustment of Input LightColorimetric temperature measurement based on CCD is strict to image exposure, so radiation image must be moderately exposured to accurately find out the actual temperature. When aperture size and shutter time are fixed, the dynamic response range o

34、f CCD spectrum is small, which leads to narrow measuring temperature range. Therefore, according to the principle of colorimetric temperature measurement and CCD imaging process, it is needed to set maximum and minimum separately for radiation image exposure to judge whether excessive exposure or un

35、derexposed. When the amount of input light is excessive or insufficient, the aperture and shutter can change automatically to make it in CCD spectrum dynamic response range for effectively expanding CCD temperature measurement range.B. Blackbody Furnace CalibrationLiterature 7 proposed that the erro

36、r caused by non-ideal impulse in CCD spectrum response characteristic can be revised by CCD response bandwidth correction coefficient K. Our study uses response bandwidth correction coefficient a and b to revise the error. The following equation can be get from equation 4:57(,410(,r r r L T K A L T

37、=ECS Transactions, 45 (33 27-36 (2013Use Wien formula into equation 5:6Similarly: 7Then:8Make:(+=1111ln5ln 2g r r gC b g r a 9 So:10Eventually, after putting equation 10 into equation 4 and reorganization the temperature T can be represented by:rg r g a G Rb C T ln 5ln 112+=11MIKRON M300 blackbody f

38、urnace is choosen as source for calibration, its temperature changes 0.5K every 8h to show its stability, emissivity is 0.999 0.0005. The noise of CCD is mainly dark current noise which is a kind of random noises 8, so the method that cauclates the average of five images shot at the same temperature

39、 and then executes mean filter is adopted to eliminate this noise 9-10. Put lens hood in front of the camera to eliminate the noise from ambient light. Choose infrared photoelectric pyrometer (973 2273K, the error is 0.5% of the reading 1K as temperature measurement instrument, its reading is consid

40、ered to be the temperature T of blackbody furnace. Between 1273 1473K, 10K is used to be the warming interval for calibration. The functional relationship between T and K is nearly index curve by observing calibration data, but directly putting index function or quadratic polynomial into the equatio

41、n 11 cant get the expression of T, therefore, equation 12 is get by75211410(exp(r r r rC K A T =75211410(exp(g g g gC K A T =2(1111lnln5ln(g rgr rgr K C K g T =+1lnrgK a bT K =+ECS Transactions, 45 (33 27-36 (2013implementing Taylor series expansion on index function and eliminating high-order terms

42、, then the function relationship between T and K can be calculated by:(Tc ba T K +=1 12The coefficient a, b and c are shown in TABLE from calibration.TABLE . Bandwidth Correction Parameter shutter time / mscorresponding temperature range/ Ka B / K c 40.11 12731323 -3.1300 0.2761103 -12.7610-4 16.79

43、13231373 -5.6485 3.3001103 -11.7610-4 8.76 13731423 -4.9325 2.8186103 -11.0910-4 4.7514231473-8.0008 6.6901103 -10.4210-4C. Target Identification on High-temperature Image with Various InterferenceHigh-temperature radiator image acquisited by CCD always carries various noises mainly cuased by ambien

44、t light, smoke scattering, and radiation light from high-temperature radiator. The former is mostly set in short-wave band, it mainly disturbs green gezer image, while the latter is mostly set in long-wave band, it mainly disturbs red gezer image. Use green gezer image segmentation to remove long-wa

45、ve band noises.Make c (x, y as the color space vector of original image coordinate (x , y , then:(=y x B y x G y x R y x ,c 13Set segmentation threshold T G for green gezer image, run equation (14 to get binary image G I , then implement point multiplication by binary image G I and originalcolor ima

46、ge c (x, y to obtain the color space vector of foreground image (G ,x y c:(=y x y x y x T y x G T y x G y x G G G G G,0,1,I c cI 14Regard red gezer image of original image as spectrum image in red band, the noises caused by ambient light is set in green band, so the segmentation of red spectrum imag

47、e can effectively remove noises from ambient light. Set segmentation threshold T R for red gezer image, then use equation 15 to get the color space vector of foregroundimage (R ,x y c :(=y x y x y x T y x R T y x R y x R R R R R,0,1,I c cI 15After the segmentation on red, green gezer image of high-t

48、emperature radiatior, if the intersection of segmentation results is taken as identification result of high-temperature radiator image, it can simultaneously eliminate the noise in long-wave and short-wave band. Use equation 16 to get color image from the primary segmentation (RG ,x y c: (=y x y x y

49、 x others T y x G T y x R y x RG RG G R RG,0,1,I c c I 且 16. Experimental ResultsUse calibrated temperature measurement system to copper melt in the mouth of blackbody furnace for measuring temperature, the standard instrument is one-time thermocouple whose value is taken as actual temperature of co

50、pper melt, the measuring temperature results are shown in TABLE .TABLE . Measuring Temperature Results after Identification on Melt Image actual temperature/ Ccalculated temperature/ Cabsolute error / Crelative error /%1202.8 1192.3 10.5 0.87 1216.1 1208.0 8.1 0.67 1209.3 1196.1 13.2 1.09From experi

51、mental results it is known that nearly 1200C the minimum relative error of measuring temperature is 0.67%, and the minimum absolute error is 8.1C.The measuring temperature results and false color display of copper melt surface whose actual temperature is 1204.2C are shown in Figure 3. Figure 3. Meas

52、uring Temperature Results and False Color DisplayFigure 3 shows that the highest temperature is 1248C, the lowest temperature is 1121C, the average temperature is 1195C. Taking the average of point temperature measured by thermocouple as actual temperature, it is deduced that the absolute error is 9

53、.2C. So experimental results show that this system can more precisely measure the temperature field of high-temperature melt, it has good practicability and is worth popularization. Conclusion(1 After discussing the structure and function of equipment needed in measuring temperature system, our study presents the condition which every equipment should satisfy, designs and builds a contactless measuring temperature system based on color CCD image sensor, introduces the hardware equipment, and develops the mea