A method and system for secondary non-uniformity correction in differential spectral filtering infrared imaging

Abstract

This invention relates to a secondary non-uniformity correction method for differential spectral filtering infrared imaging. The method comprises: based on a differential spectral filtering infrared imaging system, and building upon the two-point non-uniformity correction of a single-channel long-pass filter, adding a secondary non-uniformity correction between adjacent channels to equalize the signal response between channels. Compared with existing technologies, the secondary non-uniformity correction method for differential spectral filtering infrared imaging provided by this invention adds a secondary equalization correction on top of the two-point non-uniformity correction. This improves the uniformity of the infrared image while achieving signal response equalization between channels, enabling the infrared image to objectively reflect differences in radiation.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of infrared focal plane detector imaging and image processing, and particularly relates to a secondary non-uniformity correction method and system for differential spectral filtering infrared imaging. BACKGROUND

[0002] With the development of industrial economy, oil and natural gas has become an indispensable energy in people's life and production process, which may leak in the process of transportation, petrochemical product production and use. The leakage of industrial gas not only pollutes the environment, but also causes fire, explosion and other disasters if not handled in time, resulting in significant loss of life and property, and becoming one of the major disasters in the field of industrial production and transportation. In recent years, various infrared imaging detection modes have been proposed for industrial gas leakage detection at home and abroad. Among them, the differential spectral filtering imaging mode is a detection mode proposed by Bertin Company of France in 1991. This imaging mode adopts long-pass filtering imaging with characteristic wavelength stepping, and forms the imaging information of the waveband between the two characteristic wavelengths by subtracting the two images. Therefore, it has higher imaging signal-to-noise ratio than single-channel bandpass imaging detection, which is beneficial to the detection of gas leakage.

[0003] In an ideal case, under uniform radiation, each detection unit of an infrared focal plane array (IRFPA) detector should obtain consistent data output according to the linear response model. However, in actual cases, the output of each detection unit is inconsistent due to factors such as material and process conditions, which produces fixed pattern noise (FPN) in the image, and non-uniformity correction (NUC) is required. The non-uniformity of IRFPA reduces the image signal-to-noise ratio, blurs the details, reduces the image quality and temperature resolution of the imaging system, seriously affects the information acquisition, and greatly limits the performance of the infrared focal plane imaging system.

[0004] The two-point equalization correction algorithm is a correction process under two different filtering channel radiation imaging conditions. First, the gain correction matrix and offset correction matrix of each detection unit corresponding to the two-point correction mode are obtained by the output response of the detector in single channel, so as to ensure the uniformity of the detector unit within a certain radiation range. However, according to the working principle of the differential spectral filtering gas leakage infrared imaging system, the imaging signal is obtained by subtracting the images of two adjacent long-pass filters. Although two-point non-uniformity correction is performed for each imaging, the responses of the channels are not consistent, and further equalization between the signals of the channels is required to subtract them in the same scale and objectively reflect the difference in radiation. SUMMARY

[0005] To solve the above problems, the present application adopts the technical scheme as follows:

[0006] A secondary non-uniformity correction method for differential spectral filtering infrared imaging, the method comprising the following steps:

[0007] Based on the long-pass filter single-channel two-point non-uniformity correction of the differential spectral filtering infrared imaging system, the secondary non-uniformity correction between the adjacent two channels is added to make the signal responses between the channels equal.

[0008] Further, the secondary non-uniformity equalization correction step is as follows:

[0009] The adjacent two channels are respectively denoted as channel 1 and channel 2; it is assumed that the signal responses y1(i,j) and y2(i,j) of the channel 1 and the channel 2 are respectively

[0010] y1(i,j)=a1(i,j)x1(i,j)+b1(i,j)

[0011] y2(i,j)=a2(i,j)x2(i,j)+b2(i,j)

[0012] Wherein, x1(i,j) and x2(i,j) are the original data obtained by the detector when passing through the channel 1 and the channel 2 respectively, a1(i,j) and a2(i,j) are the gain matrices of the channel 1 and the channel 2 respectively, b1(i,j) and b2(i,j) are the offset matrices of the channel 1 and the channel 2 respectively;

[0013] The signal response of the channel 2 is made consistent with the channel 1 through linear transformation, and the linear transformation formula is:

[0014] y3(i,j)=a3(i,j)y2(i,j)+b3(i,j)=a1(i,j)x2(i,j)+b1(i,j)

[0015] Wherein, y3(i,j) represents the output data of the channel 2 after the secondary non-uniformity equalization correction;

[0016] The slope and the intercept of the curve obtained by fitting the original data x1(i,j) and the output data y1(i,j) are a1(i,j) and b1(i,j) respectively, the slope and the intercept of the curve obtained by fitting the corrected signal response y2(i,j) of the channel 2 and the output data y3(i,j) are a3(i,j) and b3(i,j) respectively, and then the values of a1(i,j), b1(i,j), a2(i,j) and b2(i,j) are used to

[0017] The following is obtained:

[0018] Further, the signal in the band-pass formed by the subtraction of the bandwidths of the two long-pass filter channels in two adjacent channels is

[0019]

[0020] Further, the two-point non-uniform correction is:

[0021]

[0022] wherein, is the true radiation estimation value of the differential spectral filtering infrared imaging system, y k (i,j) is the output value of the differential spectral filtering infrared imaging system; and are the corrected gain matrix and bias matrix of g(i,j) and o(i,j) respectively;

[0023] wherein, a(i,j) is the multiplicative non-uniform noise; b(i,j) is the additive non-uniform noise;

[0024] The corrected gain matrix and bias matrix and are substituted into to realize the two-point non-uniform correction.

[0025] Further, the calculation process of a(i,j) and b(i,j) is:

[0026] The responses of the detector to the blackbody radiation are collected at high temperature t H and low temperature t L , and the radiation values received by the detector at the two temperatures are x(t H ) and x(t L ) respectively. The responses y k (i,j,t H ) and y k (i,j,t L ) of the detector (i,j) are represented as:

[0027] y k (i,j,t H ) = a(i,j) x(t H ) + b(i,j)

[0028] y k (i,j,t L ) = a(i,j) x(t L ) + b(i,j)

[0029] According to y k (i,j,t H) and y k (i,j,t L The calculation formula of a(i,j) and b(i,j) can be obtained as follows:

[0030]

[0031]

[0032] wherein the radiation value received by the detector (i,j) is x(t H ) and x(t L The mean value of responses of all detection units is used to replace x(t

[0033]

[0034]

[0035] M×N is the size of the infrared focal plane array.

[0036] Further, the corrected gain matrix and bias matrix and are respectively:

[0037]

[0038]

[0039] Further, for the infrared focal plane array with the size of M×N, the readout signal y k (i,j) of the kth frame is:

[0040] y k (i,j) = a(i,j)×x k (i,j) + b(i,j) + n k (i,j)

[0041] wherein x k (i,j) is the infrared radiation value received by the detection element (i,j); a(i,j) is the multiplicative non-uniform noise; b(i,j) is the additive non-uniform noise; n k (i,j) is the random noise.

[0042] Further, the inverse transformation of y k (i,j) is

[0043] The application further provides a system of a secondary non-uniform correction method of differential spectral filtering infrared imaging, which comprises a differential spectral filtering infrared imaging system, a black body and a host computer; wherein the black body and the differential spectral filtering infrared imaging system are placed in parallel light paths, and the black body is used as a reference; the differential spectral filtering infrared imaging system is used for aligning the black body radiation surface and collecting the black body radiation response of each channel of the black body; the differential spectral filtering infrared imaging system is further used for adjusting the channels and completing the collection of the images of the black body; and the host computer is used for displaying and saving the images collected by the filter of the differential spectral filtering infrared imaging system.

[0044] Further, the interval between the black body and the differential spectral filtering infrared imaging system is 50 cm.

[0045] Compared with the prior art, the secondary non-uniform correction method of differential spectral filtering infrared imaging provided by the application increases the secondary equalization correction on the basis of the two-point non-uniformity correction, thereby improving the uniformity of the infrared image and realizing the equalization of the signal response between the channels, so that the infrared image can objectively reflect the difference in radiation.

[0046] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and achieved by the structure indicated in the specification, claims and drawings. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0048] Figure 1 It is a differential spectral filtering infrared imaging system filter wheel schematic diagram;

[0049] Figure 2 It is a response curve diagram of each channel changing with temperature before correction;

[0050] Fig. 3 is a two-point calibration correction principle diagram; wherein, Figure 3a It is an original response curve diagram; Figure 3b It is a curve diagram after gain correction; Figure 3c It is a curve diagram after offset correction;

[0051] Figure 4 It is a secondary non-uniform correction schematic diagram;

[0052] Fig. 5 is a comparison diagram of actual scenes before and after secondary non-uniform correction according to the present application; Figure 5a Fig. 6 shows an actual scene before correction; Figure 5b Fig. 7 shows an actual scene after correction;

[0053] Figure 6 Fig. 8 is a schematic diagram of a host computer interface. DETAILED DESCRIPTION

[0054] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0055] In the embodiments of the present application, the differential spectral filtering infrared imaging system is a self-developed instrument; the detector is purchased from Yantai Aireye Optoelectronic Technology Co., Ltd., and the model is LA6110-PL16113S00. The specific parameters of the detector are as follows: the detector material is vanadium oxide, the response wavelength band is 3-12 μm, the pixel size is 640*512, the pixel size is 17 μm, and the NETD (noise equivalent temperature difference) is less than 30 mK. The blackbody radiation surface source is selected as SR-800N-7D of CI System Co., Ltd. in Israel, which is used as a uniform background and controls the background temperature through the controller. The specific parameters of the blackbody are shown in Table 1:

[0056] Table 1 Specific parameter settings of blackbody radiation surface source

[0057] Parameter name Parameter value Emission surface size 7^”×7” Absolute temperature range 5℃-9℃ Temperature accuracy 0.007°C @ 0 < T < 50°C, 0.015°C @ 0 < T < 50°C Temperature stability ±0.003°C @ ΔT < ±10°C Set and read resolution 0.001℃ Emissivity 0.97±0.02 Non-uniformity ±0.015°C @ 1°C temperature, 80% area of central region of radiation surface

[0058] The experimental operation system in the embodiments of the present application includes a differential spectral filtering gas leakage infrared imaging system, a blackbody and its control panel, and a host computer. The secondary non-uniformity equalization correction method of the differential spectral filtering gas leakage infrared imaging provided in the embodiments of the present application uses the differential spectral filtering gas leakage infrared imaging system for experiments, the operator uses the host computer to take pictures, and the data are processed by using matlab in the later stage. The information to be counted includes: the blackbody temperature is 283-343 K (5 ℃ is adjusted each time), and the average gray value of the collected images at each temperature. The response curve before correction can be obtained by plotting the horizontal coordinate as the temperature and the vertical coordinate as the average gray value, as shown in Fig. 6. After correction, the signal responses between each channel are equalized, and the difference between the two can objectively reflect the difference of the radiation signals. Figure 2

[0059] As shown in Fig. 7, the response curve after correction is shown in Fig. 7. Figure 1 ​A differential spectral filtering infrared imaging system filter wheel is shown in the schematic diagram, Figure 1 In the formula, LP represents a high-pass filter, BP represents a band-pass filter, and SP represents a low-pass filter. The corresponding channels of the filters are as follows: channel 1 is a band-pass filter with a spectral wavelength of 6-12 nm; channel 2 is a high-pass filter with a spectral wavelength of 7490 nm; channel 3 is a high-pass filter with a spectral wavelength of 8110 nm; channel 4 is a high-pass filter with a spectral wavelength of 9000 nm; channel 5 is a high-pass filter with a spectral wavelength of 10000 nm; and channel 6 is a low-pass filter with a spectral wavelength of 11000 nm.

[0060] The correction process is as follows:

[0061] The black body and the differential spectral filtering gas leakage infrared imaging system are placed in parallel light paths at a distance of about 50 cm. The differential spectral filtering infrared imaging instrument is aligned with the black body radiation surface so that the imaging field of view is not greater than the black body surface source.

[0062] The differential spectral filtering infrared imaging instrument is started and works for a period of time (not less than 1 hour). After the temperature of the differential spectral filtering infrared thermal imager is constant, subsequent test calibration work is started.

[0063] The black body is started and set to a temperature of 10℃ (283K). After the temperature is stable, the black body radiation response of channels 1 to 6 is collected, and multiple frames of 14-bit original data are taken.

[0064] Exemplarily, the actual scene outside the window before correction is shown in FIG. 5(a). Within the working temperature range of the differential spectral filtering infrared thermal imager, the response of the IRFPA (infrared focal plane array) pixel to the infrared radiation value is approximately linear. Therefore, for an IRFPA with a size of MxN, the readout signal y k (i,j) is:

[0065] y k (i,j) = a(i,j) x k (i,j) + b(i,j) + n k (i,j) (1)

[0066] wherein x k (i,j) is the infrared radiation value received by the detection element (i,j); a(i,j) is the multiplicative non-uniform noise; b(i,j) is the additive non-uniform noise; and n k (i,j) is the random noise.

[0067] For the non-uniformity correction algorithm, the input value of the correction algorithm is the output value y k (i,j) of the thermal imaging system, and the output value is the real radiation estimate value of the thermal imaging system. If the influence of random noise is ignored, the inverse transform of equation (1) is the process of non-uniformity correction, and the response before correction is shown in Figure 3(a):

[0068]

[0069] in,

[0070] The true values ​​of g(i,j) and o(i,j) are unknown; the non-uniformity correction algorithm obtains estimates of these two correction factors. and and These are the correction gain matrix and bias matrix. Non-uniformity correction models are divided into one-point correction models and two-point correction models. The correction process is shown in equation (3):

[0071]

[0072] Set the temperature to 15℃, 20℃, ..., 70℃ in sequence (5℃ interval each time), and repeat step 3;

[0073] The raw data from 10℃ to 70℃ obtained from channel 1 were averaged.

[0074] A straight line is fitted to the mean blackbody radiation response from 10℃ to 70℃ calculated for channel 1, as follows: Figure 2 The figure shows the response fitting curves of each channel as a function of temperature before correction. A suitable high temperature point t was selected for fitting. H and low temperature point t L The detector's response to blackbody radiation was collected at the following temperatures; the radiation values ​​received by the detector at these two temperatures were x(t) and x(t) respectively. H ) and x(t L The response y of detector (i,j) k (i,j,t H ) and y k (i,j,t L ) is represented as:

[0075] y k (i,j,t H )=a(i,j)×x(t H )+b(i,j) (4)

[0076] y k (i,j,t L )=a(i,j)×x(t L )+b(i,j) (5)

[0077] The multiplicative factor a(i,j) and the additive factor b(i,j) of the (i,j) detector element can be obtained by combining equation (4) and equation (5) as shown in equation (6) and equation (7) respectively:

[0078]

[0079]

[0080] where x(t H ) and x(t L ) are replaced by the mean value of all detector responses, i.e. x(t H ) and x(t L ) are shown in equation (8) and equation (9) respectively:

[0081]

[0082]

[0083] According to the two-point correction algorithm, the gain matrix and the offset matrix of channel 1 are calculated, and the results of gain correction and offset correction are shown in (b) and (c) of FIG. 3; the calculation formulas of the gain matrix and the offset matrix of channel 1 are shown in equation (10) and equation (11) respectively:

[0084]

[0085]

[0086] The experimenter changes the channel by manually rotating the filter wheel one by one, repeats steps 5-7, and obtains the gain matrix and the offset matrix of channels 2-6 respectively, and then the gain matrix and the offset matrix of each channel are respectively brought into the correction formula (3), i.e. the two-point correction is completed.

[0087] According to the existing classical theoretical model (CTM), the simulation results show that the narrower the bandwidth, the lower the gray value, which is seriously inconsistent with the actual situation. According to the working principle of the differential spectral filtering gas leakage infrared imaging system, the imaging signal is obtained by subtracting the images of two adjacent long-pass filters, although two-point non-uniformity correction is performed each time, but due to the inconsistent response of each channel, the signal response between channels needs to be balanced, and the difference between the two can objectively reflect the difference of the radiation signal, so secondary equalization correction is performed on the basis of two-point calibration correction.

[0088] The values of a1, a2, b1 and b2 in equation (12) can be obtained by fitting the curve after two-point correction, as shown in equation (13) and equation (14): Figure 4The linear transformation parameters of each channel response consistent with channel 1 can be obtained according to formula (13), that is, the secondary non-uniformity equalization correction is completed.

[0089] Exemplarily, the process of the secondary non-uniformity equalization correction is as follows: assuming that the responses of channel 1 and channel 2 are y1(i,j) and y2(i,j) in formula (12) respectively:

[0090] y1(i,j)=a1(i,j)x1(i,j)+b1(i,j)

[0091] y2(i,j)=a2(i,j)x2(i,j)+b2(i,j) (12)

[0092] In order to make the response of channel 2 consistent with channel 1, linear transformation is required, as shown in formula (13):

[0093] y3(i,j)=a3(i,j)y2(i,j)+b3(i,j)=a1(i,j)x2(i,j)+b1(i,j) (13)

[0094] Wherein,

[0095] Similarly, the linear conversion formula of the slope and intercept when channels 2, 3, 4, 5 and 6 are consistent with the previous channel can be obtained, and formula (13) is brought into the formula to complete the secondary non-uniformity equalization correction.

[0096] Therefore, the band-pass internal signal composed of two long-pass filters is

[0097]

[0098] The filter wheel of the differential spectral filtering gas leakage infrared imaging system is used for adjusting the channels, the black body is imaged through different filter channels, and the collected images are transmitted to the upper computer for display and storage (the schematic diagram of the upper computer is shown in FIG. 2), so that subsequent processing is performed, and the actual scene after correction is shown in FIG. 5(b). Figure 6

[0099] Under ideal conditions, the response output of each detection unit of the infrared focal plane array under the radiation of the external uniform thermal field should be completely consistent, and actually, due to the influence of non-uniformity of semiconductor materials for manufacturing devices, flooding errors, defects, process conditions and the like, the response output of each detection unit is not completely consistent, which is the non-uniformity (nonuniformity, NU) of the infrared focal plane array. The present application uses the definition of the non-uniformity of the infrared focal plane array in the “Technical Specification for Testing Parameters of Infrared Focal Plane Array” of the People's Republic of China in 1999, and the definition formula is as follows: ​

[0100]

[0101] In the formula:

[0102]

[0103] In formula (15), V ij is the response output voltage of the detection unit corresponding to the i-th row and j-th column of the infrared focal plane array, V oavg is the average value of the response output voltage signals of all effective detection units on the infrared focal plane array, M and N are the number of rows and columns of the infrared focal plane array respectively, d is the number of dead detection units in the infrared focal plane array, h is the number of overheated detection units in the infrared focal plane array, and the dead detection units and the overheated detection units are collectively referred to as invalid detection units.

[0104] Therefore, according to the above formula (15) and formula (16), the NU before the secondary non-uniformity correction is calculated to be 0.06327, and the NU after the secondary non-uniformity correction is calculated to be 0.002636.

[0105] In summary, the application provides a secondary non-uniformity equalization correction method for differential spectral filtering infrared imaging, which increases the secondary equalization correction on the basis of two-point non-uniformity correction, improves the uniformity of the infrared image, and realizes the equalization of the signal response between channels, so as to objectively reflect the difference in radiation.

[0106] Although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application.

Claims

1. A method for quadratic non-uniform correction of differential spectral filtering infrared imaging, characterized in that, The method comprises the following steps: Based on the long-pass filter single-channel two-point non-uniformity correction of the differential spectral filtering infrared imaging system, the two-point non-uniformity correction between adjacent two channels is added to make the signal response between the channels uniform; The two-point non-uniformity correction is as follows: The two adjacent channels are denoted as channel 1 and channel 2, respectively; it is assumed that the signal responses of channel 1 and channel 2 and are wherein, and are the raw data acquired by the detector while passing through channel 1 and channel 2, respectively, are the gain matrices of channel 1 and channel 2, respectively, are the bias matrices of channel 1 and channel 2, respectively. The signal response of channel 2 is made consistent with that of channel 1 through linear transformation, and the linear transformation formula is as follows: wherein, represents the output data of channel 2 after the second non-uniformity equalization correction; The original data and output data are fitted to obtain the slope and intercept of the curve respectively and , the corrected signal response of channel 2 and output data are fitted to obtain the slope and intercept of the curve respectively and , and then the values of , , , are obtained. obtained .

2. The method of claim 1, wherein, The signal in the band-pass formed by the subtraction of the bandwidths of the two long-pass filters in the adjacent two channels is as follows: 。 3. The method of claim 1, wherein, The two-point non-uniformity correction is as follows: wherein, is a true radiance estimate for a differential spectral filter infrared imaging system, y k ( i , j ) is an output value for a differential spectral filter infrared imaging system; and are respectively and corrected gain matrix and bias matrix; , wherein is multiplicative non-uniform noise; is additive non-uniform noise.

4. The method of claim 3, wherein, and The calculation is: The response of the detector to blackbody radiation is measured at high temperature and low temperature , and the values of the radiation received by the detector at these two temperatures are and respectively. The response of the detector at these two temperatures is given by and respectively. According to and the calculation formula can be obtained and respectively: ； wherein the detector The received radiation values are and replaced by the mean value of the responses of all detection units, i.e.: ； is the size of the infrared focal plane array.

5. The method of claim 4, wherein, corrected gain matrix and bias matrix and are respectively: 。 6. The method of claim 4, wherein, For a size of An infrared focal plane array, the readout signal of the kth frame of which is is: wherein, is a detection element received infrared radiation value; is multiplicative non-uniform noise; is additive non-uniform noise; is random noise.

7. The method of claim 6, wherein, The inverse transform of .

8. A system for implementing the two-point non-uniformity correction method of the differential spectral filtering infrared imaging according to any one of claims 1-7, characterized in that, The system comprises a differential spectral filtering infrared imaging system, a black body and a host computer; The black body and the differential spectral filtering infrared imaging system are placed in parallel light paths, The black body is used as a reference; The differential spectral filtering infrared imaging system is used to align the black body radiation surface, and collect the black body radiation response of each channel of the black body; The differential spectral filtering infrared imaging system is also used to adjust the channels and complete the collection of the images of the black body; The host computer is used to display and save the images collected by the filter of the differential spectral filtering infrared imaging system.

9. The system of claim 8, wherein, The interval between the black body and the differential spectral filtering infrared imaging system is 50 cm.

Images