Wide-range temperature measurement method based on adjustable normalization algorithm

Abstract

The invention provides a wide-range temperature measurement method based on an adjustable normalization algorithm, and the method comprises the following steps: 1) employing a blackbody furnace and a silicon-based image sensor to calibrate a radiation temperature measurement system, and building a linear relation calibration formula between an input radiation intensity M and an output gray value G; 2) in actual temperature measurement, obtaining an effective gray value G (t0) of the silicon-based image sensor under the exposure time t0 through an adjustable normalization algorithm; 3) substituting the gray value G (t0) obtained in the step 2) into the linear relation calibration formula in the step 1), and calculating the radiance M of the object to be measured; and 4) according to inverse function operation of the Planck formula, the temperature T of the to-be-measured object is solved by the radiation intensity M of the to-be-measured object. Aiming at the problems of narrow temperature measurement range and low temperature measurement upper limit caused by output saturation of the existing silicon-based image sensor in high-temperature measurement, wide-range temperature measurement is realized by dynamically adjusting exposure time and combining linear calibration and Planck formula inverse function operation.

Claims

1. A wide range temperature measurement method based on an adjustable normalization algorithm, characterized in that: It includes the following steps: 1) Calibrate the radiation temperature measurement system using a blackbody furnace and a silicon-based image sensor. Establish a linear relationship calibration formula between the input radiance M and the output gray value G at a fixed exposure time t0: M = K·G + b (t = t0); Where, K is the system constant, b is the dark signal, and t is the exposure time; 2) In actual temperature measurement, the effective grayscale value G of the silicon-based image sensor at exposure time t0 is obtained through an adjustable normalization algorithm. （t0） ; 3) The gray value G obtained in step 2) （t0） Substitute the linear relationship calibration formula in step 1) to calculate the radiometry M of the object under test; 4) According to the inverse function operation of Planck's formula, solve the temperature T of the object to be measured from the radiance M of the object to be measured.

2. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 1, characterized in that: Step 2) specifically includes the following steps: 2.1) Set the number of iterations i=0, t=t0, and obtain the current grayscale value G (t) ; 2.2) The gray value G （t） and grayscale saturation threshold G （m） Compare, if 0< G (t) < G m , then it is judged that the silicon-based image sensor is not saturated, G （t0） =G （t） , and go to step 2.8); if G (t) ≥ G m , then the silicon-based image sensor is judged to be saturated and the process goes to step 2.3); 2.3) Adjust t=t / a and re-obtain the current grayscale value G （t） , a is the first exposure time adjustment coefficient, a>1; 2.4) The current grayscale value G will be re-obtained （t） and threshold G （m） Compare, if 0< G (t) < G m , then it is determined that the silicon-based image sensor is not saturated, and proceed to step 2.5). If G (t) ≥G m , it is determined that the silicon-based image sensor is saturated, and the process returns to step 2.3); 2.5) Get the current grayscale value G （t） , and let G （ti） = G （t） , t i =t, i=i+1; 2.6) If i < n, adjust the exposure time t = t×c, and return to step 2.5); if i ≥ n, enter step 2.7). c is the second exposure time adjustment coefficient, and c < 1; 2.7) Obtain n images at different exposure times t i The unsaturated grayscale value output G （ti） , based on the least squares method to fit the gray value G (ti) With exposure time t i Linear relationship, substitute t=t0 into the gray value G (ti) With exposure time t i The effective gray value G is obtained by the linear relationship (t0) ; 2.8) Output effective grayscale value G (t0) .

3. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 2, characterized in that: Fitting the gray value G based on the least squares method (ti) With exposure time t i The linear relationship is as follows: Based on n different exposure times t i The unsaturated grayscale value output G （ti） , get the gray value vector and exposure time vector Based on the least squares method, the functional relationship between the grayscale value G and the exposure time t is obtained from the two vectors: G=p×t+q; Substituting t=t0 into the effective grayscale value G (t0) ; Where p is the slope, which represents the change in grayscale value per unit exposure time; q is the intercept, which represents the comprehensive offset of the silicon-based image sensor.

4. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 2, characterized in that: In step 2.6), n is the number of acquired unsaturated gray values, satisfying 10 ≤ n ≤ 30; Furthermore, during the iteration process in step 2.6), when the exposure time t is adjusted with a preset step length or ratio, the exposure time t after each adjustment must satisfy the following requirements: (ti) The value is within the linear region of the grayscale value output of the silicon-based image sensor.

5. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 4, characterized in that: The preset step parameter c of exposure time adjustment satisfies 1 / 2<c<1, and the exposure interval t after each adjustment is i =t i-1 ×c corresponds to the gray value G (ti) Both are within the linear response region of the sensor.

6. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 1, characterized in that: In step 4), the method of inverse function operation through Planck's formula includes: Planck's formula: , where M is the radiometric value of the object to be measured; λ is the measurement wavelength; T is the thermodynamic temperature of the object to be measured; C1 is the first radiation constant, C 1 = 3.7415×10 -16 (w·m 2 ） ; C2 is the second radiation constant, C 2 = 1.439×10 -2 (w·m 2 ） ;ε(λ,T) is the object emissivity; Inverse function operation: T=f -1 [M(λ, T)], directly substitute the obtained radiance M to obtain T.

7. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 1, characterized in that: The calibration method using a blackbody furnace in step 1) includes: Generate at least five standard radiation sources with different temperatures through the blackbody furnace, the temperature range covers 800K to 1800K, and the adjacent temperature interval is not less than 200K; Based on the calibration data in different temperature intervals, establish a piecewise linear relationship. Among them, 800K–1000K, 1000K–1300K, and 1300K–1800K respectively correspond to independent calibration parameters to adapt to the response characteristics of the silicon-based sensor in different temperature regions.

8. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 7, characterized in that: The establishment of the piecewise linear relationship includes the following steps: Generate at least two standard temperature points for calibration through the blackbody furnace within each temperature interval; within each temperature interval, adjust the calibration exposure time so that the gray value output of the silicon-based image sensor is located in the 60%–90% of its linear response region, and independently calculate the corresponding calibration parameters K and b for this interval to form a piecewise linear mapping relationship; At the junctions 1000K and 1300K of adjacent temperature intervals, use the weighted average method to interpolate the calibration parameters of adjacent intervals to eliminate the measurement error introduced by temperature jumps.

9. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 8, characterized in that: The method of interpolating the calibration parameters of adjacent intervals using the weighted average method is: According to the relative distance between the temperature T to be measured and the junction temperature, calculate the weights of the calibration parameters of adjacent intervals adjustable. The weight value is inversely proportional to the temperature distance; Based on the weights, the calibration parameters K and b of adjacent intervals are weighted averaged to generate a smooth transition fusion parameter K fuse and b fuse , used for radiometry calculations near the junction temperature.

10. The wide-range temperature measurement method based on an adjustable normalization algorithm according to claim 9, characterized in that: The weight distribution and parameter fusion rules of weighted average include the following steps: s Within the preset transition range of ±20K at the temperature interval junctions 1000K and 1300K, according to the difference between the temperature to be measured and the junction temperature, distribute the weights of the calibration parameters of adjacent temperature regions in a linear proportion; When the temperature to be measured is at the junction temperature, the parameter weights of both sides of the temperature region each account for 50%; As the temperature shifts to any temperature region, the weight of the corresponding temperature region increases proportionally, and the weight of the other temperature region decreases; Perform weighted calculation on the calibration parameters K and b of adjacent temperature regions according to the allocated weights.

Images