An efficient non-uniform correction method considering the effects of attenuator and integration time
By establishing a non-uniform correction model that considers attenuators and integration time, and using weighted least squares and weighted linear least squares methods to solve the non-uniform correction coefficients, the problems of cumbersome and time-consuming correction and limited storage resources in infrared imaging systems are solved, and efficient non-uniform correction is achieved.
Patent Information
- Application Number
- CN202410953054.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-16
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2044-07-16
AI Technical Summary
Existing infrared imaging systems using variable attenuators and variable integration times employ cumbersome and time-consuming non-uniform correction methods, which also result in limited storage resources and make it difficult to meet the needs of spaceborne and airborne systems.
By establishing a non-uniform correction model that considers the attenuator and integration time, the non-uniform correction coefficients are solved using the weighted least squares method and the weighted linear least squares method, thereby reducing the number of correction coefficients and improving efficiency.
It achieves efficient non-uniform correction, reduces storage resource requirements, simplifies operation procedures, and improves correction performance.
Smart Images

Figure CN118822914B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of infrared image non-uniformity correction, specifically relating to an efficient non-uniformity correction method that considers the effects of attenuator and integration time. Background Technology
[0002] Due to limitations in the manufacturing processes of core components in current infrared imaging systems, infrared images exhibit non-uniformity, manifesting as stripe noise, block noise, speckle noise, and halo effects. These defects severely impact the image quality, necessitating non-uniformity correction. Currently, to detect various targets, infrared imaging systems typically employ attenuators and integration times to achieve wide dynamic range detection. However, the two-point non-uniformity correction method used in engineering demonstrates its drawbacks. The variable attenuator and integration time result in numerous correction levels, requiring correction for each level, making the process cumbersome and time-consuming. Furthermore, the increased number of correction levels necessitates storing a larger number of correction coefficients, which is insufficient for spaceborne and airborne systems with limited storage resources. Additionally, achieving superior non-uniformity correction sometimes requires multiple segments of two-point non-uniformity correction for each level to adapt to scenarios with significant background radiation variations. This further increases the number of two-point correction coefficients, exacerbating the drawbacks of two-point correction. Summary of the Invention
[0003] The purpose of this invention is to provide an efficient non-uniform correction method that takes into account the effects of attenuation plates and integration time, in order to solve or at least alleviate the problems mentioned above.
[0004] The technical solution adopted in this invention is: an efficient non-uniform correction method considering the effects of attenuator and integration time, the method comprising: Step 1: Establish a non-uniform correction model that considers the attenuator and integration time through the response model of the infrared imaging system with variable attenuator and integration time; Step 2: Execute the blackbody image acquisition strategy based on the linear temperature range of each integration time setting and attenuator setting. Step 3: Based on the non-uniform correction model considering the attenuator and integration time in Step 1, establish the objective function using the weighted least squares method, and use the optimization algorithm to solve for the eight non-uniform correction coefficients of the first few pixels of the image. The eight non-uniform correction coefficients include: correction slope, correction intercept related to stray radiation of the infrared imaging system before the attenuator, correction intercept related to stray radiation of the infrared imaging system after the attenuator, quadratic correction term related to integration time, cubic correction term related to integration time, correction intercept related to the attenuator's own radiation, correction intercept related to the detector's dark current, and coefficient related to the attenuator's emissivity. Step 4: Substitute one of the non-uniform correction coefficients obtained in Step 3, namely the coefficient related to the attenuator emissivity, into the non-uniform correction model considering the attenuator and integration time in Step 1. Linearize the non-uniform correction model and use the weighted linear least squares method to solve for the remaining seven non-uniform correction coefficients of the remaining pixels.
[0005] Beneficial effects of this invention: (1) Time-saving and efficient: Establish an efficient non-uniform correction model that considers the effects of attenuation plate and integration time. The advantage of this model is that efficient non-uniform correction can be achieved using only a few blackbody temperature correction points. (2) Greatly reduces the number of correction coefficients to be stored: The entire range uses a single correction equation, requiring only a few correction coefficients, and their number is independent of the number of attenuators and integration time ranges. In contrast, the number of correction coefficients required by the traditional non-uniform correction method is linearly related to the number of attenuators and integration time ranges. Therefore, this method can greatly reduce the number of non-uniform correction coefficients in the variable attenuator and integration time infrared imaging system. Attached Figure Description
[0006] Figure 1 A flowchart is shown for an efficient non-uniformity correction method that takes into account the effects of attenuator and integration time according to an embodiment of the present invention. Figure 2(a) shows the original image of a scene acquired by a 100% attenuation filter and an integration time of 0.5ms according to an embodiment of the present invention; Figure 2(b) shows a two-point non-uniform correction image of a scene acquired with a 100% attenuation filter and an integration time of 0.5ms according to an embodiment of the present invention. Figure 2(c) shows the corrected image of the scene acquired by the present invention using a 100% attenuation filter and an integration time of 0.5ms according to an embodiment of the present invention; Figure 3(a) shows the original image of a scene acquired by a 10% attenuation filter and an integration time of 0.1 ms according to an embodiment of the present invention; Figure 3(b) shows a two-point non-uniform correction image of a scene acquired with a 10% attenuation filter and an integration time of 0.1 ms according to an embodiment of the present invention. Figure 3(c) shows the corrected image of the scene acquired by the present invention using a 10% attenuation filter and an integration time of 0.1 ms according to an embodiment of the present invention; Figure 4(a) shows the original image of a scene acquired by a 5% attenuation filter and an integration time of 0.1 ms according to an embodiment of the present invention; Figure 4(b) shows a two-point non-uniform correction image of a scene acquired with a 5% attenuation filter and an integration time of 0.1 ms according to an embodiment of the present invention. Figure 4(c) shows the corrected image of the scene acquired by the present invention using a 5% attenuation filter and an integration time of 0.1 ms according to an embodiment of the present invention. Detailed Implementation
[0007] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
[0008] The technical solution of this invention first involves establishing a non-uniform correction model considering the attenuator and integration time through a response model of an infrared imaging system with a variable attenuator and integration time. Then, based on this model, an efficient non-uniform correction method considering the influence of the attenuator and integration time is designed, including a blackbody image acquisition strategy. Based on the non-uniform correction model considering the attenuator and integration time, a weighted least squares method is used to establish an objective function. The non-uniform correction coefficients of the first few pixels of the image are solved using an optimization algorithm based on the acquired blackbody image. The non-uniform correction coefficients σ of the first few pixels are substituted into the non-uniform correction model considering the attenuator and integration time to linearize it. Finally, the non-uniform correction coefficients of the remaining pixels are solved using a weighted linear least squares method.
[0009] Figure 1 A flowchart illustrating an efficient non-uniformity correction method considering the effects of attenuator and integration time according to an embodiment of the present invention is shown below. Figure 1 A highly efficient non-uniformity correction method according to embodiments of the present invention is described. For example... Figure 1 As shown, the method includes: Step 1: Establish a non-uniform correction model that considers the attenuator and integration time through the response model of the infrared imaging system with variable attenuator and integration time.
[0010] The response model of the variable attenuator and integration time infrared imaging system is shown in equation (1): (1) In the formula, DN is the mean of the image, and t is the integration time. Let L be the system response gain per unit time, and L be the target radiation. The transmittance of the attenuation film, The bias is caused by stray radiation from the infrared imaging system in front of the attenuator. The bias caused by stray radiation from the infrared imaging system after the attenuator. The bias is caused by the attenuator's own radiation. It is the grayscale response caused by the dark current of the detector. This represents the functional relationship between the actual transmittance of the attenuator and its design value. It is the coefficient of the quadratic term in the functional relationship between the actual transmittance of the attenuator and the design value. It is the coefficient of the first-order term in the functional relationship between the actual transmittance of the attenuator and the design value. σ is a constant term in the functional relationship between the actual transmittance of the attenuator and its design value, where σ is a coefficient related to the attenuator's emissivity. This is an empirical relationship for the emissivity of the attenuator.
[0011] The purpose of non-uniform correction is to correct the gray values of each pixel to the image mean. Based on the response model of a variable attenuator and an integral-time infrared imaging system, the image mean can be written as... (2) in, It is the average gray level of the image. It is the mean of the image response gain. The biased mean is caused by stray radiation from the infrared imaging system in front of the attenuator. It is the bias mean caused by stray radiation from the infrared imaging system after the attenuator. It is the bias mean caused by the attenuator's own radiation. It is the average grayscale response caused by the dark current of the detector.
[0012] The gray value of the pixel at row i and column j can be written as: (3) in, Let be the grayscale value of pixel i and j in row i and column j, where i is the row of the image and j is the column of the image.
[0013] Combining equations (2) and (3), the gray value of the pixel at coordinates (i, j) can be written as: (4) Equation (4) can be simplified to: (5) The meanings of each parameter are as described in formula (6).
[0014] To correct the non-strict proportional relationship between integration time and response, the integration time is incorporated into the quadratic and cubic correction coefficients. The final mathematical model for wide dynamic range non-uniformity correction is then: (6) Equation (6) shows the mathematical model for wide dynamic range non-uniform correction considering integration time and attenuator, where k is the correction slope. This is the correction intercept related to stray radiation from the infrared imaging system in front of the attenuator. The subscript (i, j) indicates this parameter at coordinates (i, j). It is the correction intercept related to stray radiation in the infrared imaging system after the attenuator. It is the correction intercept related to the attenuator's own radiation. It is the correction intercept related to the detector's dark current. It is the correction intercept related to stray radiation in the infrared imaging system after the attenuator. It is a quadratic correction term related to the integration time. It is a three-stage correction term related to the integration time.
[0015] Step 2: Based on the linear temperature range of each integration time setting and attenuator setting, execute the blackbody image acquisition strategy. For each attenuator setting, select one temperature point to maximize the number of integration time settings included, and then select one or two more temperature points to supplement attenuator settings with larger calibration temperature ranges, ensuring that the selected temperatures cover the low-temperature range. In one implementation, at least eight blackbody images are acquired.
[0016] In one example, an infrared imaging aperture of 450mm was used, employing a mid-wave band of 3.7~4.8μm, a mercury cadmium telluride detector with 640×512 pixels and a pixel size of 15μm, and Stirling cooling was used. Integration time settings were available: 0.1ms, 0.25ms, 0.5ms, 1ms, 1.5ms, and 3ms. Three neutral density attenuators with transmittance settings of 100%, 10%, and 5% were designed in the optical path in front of the detector, and the attenuator settings were switched by rotating a target wheel. The blackbody used was the HgH ECN 100 N20 model, a large area blackbody with an effective radiating area of 500×500mm. The linear temperature ranges for each integration time and attenuator setting of the infrared imaging system are shown in Table 1. For each attenuator setting, a temperature point is selected to maximize the number of integration time settings included. Then, one or two more temperature points are selected to supplement attenuator settings with larger calibration temperature ranges, ensuring that the selected temperatures cover the low-temperature range. The required effective data volume is no less than eight. Selecting a temperature point to maximize the number of integration time settings under each attenuator setting means that each attenuator setting corresponds to a temperature range, and each temperature point within that range corresponds to one or more integration time settings. Selecting a temperature point that corresponds to the most integration time settings is crucial. For some attenuator settings with larger temperature ranges, selecting one or two more temperature points to supplement those settings means that some attenuator settings correspond to a large temperature range, requiring the selection of one or two additional temperature points to constrain the attenuator setting and achieve better non-uniformity correction. In this field, the low-temperature range generally refers to the low-temperature segment of the temperature range corresponding to the attenuator setting. Effective data volume refers to the data that can participate in the non-uniformity correction calculation.
[0017] Table 1
[0018] For each attenuator setting, a temperature point is selected to maximize the number of integration time settings. For example, 50℃ can be selected for the 100% attenuation setting, 110℃ for the 10% attenuation setting, and 170℃ for the 5% attenuation setting. To improve the accuracy of non-uniform correction in the low radiation range, an 80℃ temperature point is added to cover the low-temperature range of both the 10% and 5% attenuators. Therefore, only 4 blackbody temperature points are needed to complete the non-uniform correction of the entire infrared imaging system.
[0019] Step 3: Based on the non-uniform correction model considering attenuator and integration time in Step 1, establish the objective function using weighted least squares method, and use the blackbody image acquired in Step 2 to solve for the non-uniform correction coefficients of the first few pixels of the blackbody image using an optimization algorithm.
[0020] The weighted least squares method can take into account the non-uniformity correction effect of images with small gray values. The objective function for solving the correction coefficients using the weighted least squares method is as follows: (7) The symbols in the formula have the same physical meaning as those in formula (6), m is the m-th blackbody image, n indicates that there are n blackbody images in total, and J is the objective function.
[0021] To minimize the objective function J of the first few pixels of a blackbody image, optimization algorithms include the steepest gradient method, the quasi-Newton method, the conjugate gradient method, and the interior point method. Minimizing the objective function J yields eight unknown parameters, namely the non-uniform correction coefficients and the correction slope. Correction intercept related to stray radiation from infrared imaging systems before attenuation filter Correction intercept related to stray radiation in infrared imaging systems after attenuation filter Secondary correction terms related to integration time Three correction terms related to integration time Correction intercept related to the attenuator's own radiation Correction intercept related to detector dark current And the coefficient σ related to the emissivity of the attenuator. Here, "the first few pixels" refers to 5-10 pixels, and the principle for determining the number is that the solved σ is stable and changes little.
[0022] Step 4: Substitute one of the non-uniform correction coefficients obtained in Step 3, namely the coefficient σ related to the attenuator emissivity, into the non-uniform correction model considering the attenuator and integration time in Step 1, thereby linearizing the non-uniform correction model. Then, use the weighted linear least squares method to solve for the non-uniform correction coefficients of the remaining pixels.
[0023] For all pixels, the coefficient σ related to the attenuator emissivity is the same. Thus, by substituting the σ calculated for the first few pixels into the non-uniformity correction model, the model can be linearized, thereby reducing a lot of computation. Only step 4 is needed to solve for the remaining 7 non-uniformity correction coefficients for the remaining pixels.
[0024] According to the weighted linear least squares theory, the remaining unknown non-uniformity correction coefficients for the remaining pixels are calculated as follows: (8) in p = [ k i , j b 1 , i , j b 21 , i , j b 22 , i , j b 23 , i , j b 3 , i , j b 4 , i , j ] , X = [ DN 1 ¯ t 1 ⋅ t 1 t 1 t 2 1 t 3 1 ( 1 − t 1 s ) ⋅ t 1 1 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ DN n ¯ t n ⋅ t n t n ⋅ t n t n t 2 n t 2 n 1 ] , y = [ DN 1 , i , j ⋮ DN n , i , j ] , W = [ 1 DN 2 1 , i , j 0 0 0 ⋱ 0 0 0 1 DN 2 n , i , i ] , The symbols in the formula have the same physical meaning as those in formula (1), the subscripts 1 to n represent the number of the blackbody image, X is the coefficient matrix related to the 8 non-uniform correction coefficients, y is the gray vector of pixel i and j, and W is the weight coefficient matrix.
[0025] The scene images were acquired, and the non-uniform correction processing results using the method of this invention and the two-point non-uniform correction method are shown in Figures 2(a), 2(b), 2(c), 3(a), 3(b), 3(c), 4(a), 4(b), and 4(c).
[0026] It is evident that for an infrared imaging system with 3 attenuators and 6 integration time settings, this invention achieves the same excellent correction performance as the two-point non-uniform correction method, which is maturely used in engineering. In one embodiment of this invention, 4 blackbody temperature points efficiently and quickly complete the non-uniform correction of all attenuator settings and integration time settings, while the two-point non-uniform correction method requires selecting two temperature points for correction at each setting, which is time-consuming and labor-intensive. Furthermore, for 3 attenuators and 6 integration time settings, the two-point non-uniform correction method requires storing 36 correction coefficients, while the method in this paper only requires 8 correction coefficients (and 1 parameter is shared by all pixels), reducing the storage resource requirement by about 5 times.
[0027] The techniques known in the art involved in this invention are not described in detail. Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.
[0028] Although the invention has been described with respect to a limited number of embodiments, those skilled in the art will understand from the foregoing description that other embodiments are conceivable within the scope of the invention described herein. Furthermore, it should be noted that the language used in this specification has been chosen primarily for readability and instructional purposes, and not for the purpose of explaining or limiting the subject matter of the invention.
Claims
1. A highly efficient non-uniform correction method considering the effects of attenuation plates and integration time, characterized in that, Includes the following steps: Step 1: Establish a non-uniform correction model that considers the attenuator and integration time through the response model of the infrared imaging system with variable attenuator and integration time; Step 2: Execute the blackbody image acquisition strategy based on the linear temperature range of each integration time setting and attenuator setting. Step 3: Based on the non-uniform correction model considering the attenuator and integration time in Step 1, establish the objective function using the weighted least squares method, and use the optimization algorithm to solve for the eight non-uniform correction coefficients of the first few pixels of the image. The eight non-uniform correction coefficients include: correction slope, correction intercept related to stray radiation of the infrared imaging system before the attenuator, correction intercept related to stray radiation of the infrared imaging system after the attenuator, quadratic correction term related to integration time, cubic correction term related to integration time, correction intercept related to the attenuator's own radiation, correction intercept related to the detector's dark current, and coefficient related to the attenuator's emissivity. Step 4: Substitute one of the non-uniform correction coefficients obtained in Step 3, namely the coefficient related to the attenuator emissivity, into the non-uniform correction model considering the attenuator and integration time in Step 1. Linearize the non-uniform correction model and use the weighted linear least squares method to solve the remaining seven non-uniform correction coefficients for the remaining pixels. Step 1 includes: The non-uniform correction model considering the attenuator and integration time is as follows: (6) Where k is the correction slope, and b1 is the correction intercept related to stray radiation from the infrared imaging system in front of the attenuator. It is the correction intercept related to stray radiation in the infrared imaging system after the attenuator. This is a quadratic correction term related to the integration time. For the third correction term related to integration time, It is the correction intercept related to the attenuator's own radiation. It is the correction intercept related to the detector's dark current, where i is the row of the image and j is the column of the image. For the gray values of pixels in row i and column j, denoted as the mean of the image, t as the integration time, τ as the transmittance of the attenuator, and σ as the coefficient related to the emissivity of the attenuator.
2. The efficient non-uniform correction method considering the effects of attenuator and integration time as described in claim 1, characterized in that, Step 2 includes: For each attenuator setting, select one temperature point to maximize the number of integration time settings. Then, select one or two more temperature points to supplement the attenuator settings with a large calibration temperature range, ensuring that the selected temperatures cover the low-temperature range. Collect at least eight blackbody images.
3. The efficient non-uniform correction method considering the effects of attenuator and integration time as described in claim 1, characterized in that, Step 3 includes: Based on the non-uniform correction model considering the attenuator and integration time in step 1, the objective function is established using the weighted least squares method, including: (7) Where m is the m-th blackbody image, n represents the total number of blackbody images, and J is the objective function; by minimizing the objective function J for the first few pixels, eight non-uniform correction coefficients are obtained.
4. The efficient non-uniform correction method considering the effects of attenuator and integration time as described in claim 1, characterized in that, In step 4, the weighted linear least squares method is used to solve for the remaining seven non-uniformity correction coefficients of the remaining pixels, including: (8) in , , , , Wherein, subscripts 1 to n represent the numbers of the blackbody image, X is the coefficient matrix related to the 8 non-uniform correction coefficients, y is the gray-level vector of pixels i and j, and W is the weight coefficient matrix.
Citation Information
Patent Citations
Visible light auxiliary multiband infrared real-time imaging analysis system
CN105737989A
KR20230033302A
