A method for correcting non-uniformity of an infrared linear array detector

Abstract

The application provides an infrared line detector non-uniformity correction method and device, the method comprises the following steps: intercepting part of the infrared line circumferential scanning image from the original infrared line circumferential scanning image; using a guided filter model to denoise the selected infrared line circumferential scanning image; using a linear fitting model based on the guided filter to calculate a linear correction coefficient; using the obtained linear correction coefficient to correct the non-uniformity of the whole frame of infrared line circumferential scanning image. The infrared line detector non-uniformity correction method has high real-time performance and good denoising effect.

Description

Technical Field

[0001] This invention relates to the field of digital image processing technology, and more specifically, to a method and apparatus for correcting non-uniformity of an infrared linear detector. Background Technology

[0002] In recent years, infrared imaging systems have seen significant improvements in both technology and manufacturing processes, leading to their widespread application in thermal imaging, fire detection, and aerial reconnaissance of small targets. Common infrared detectors are categorized into external infrared array detectors and linear infrared circular scan detectors. However, during infrared imaging, the detector's response coefficient changes due to variations in ambient temperature and imaging circuit voltage, resulting in striped, non-uniform noise in the infrared system image. This type of noise is generally difficult to remove directly in hardware; common solutions fall into two categories: calibration-based correction methods and scene-based correction methods.

[0003] Calibration-based correction methods typically require acquiring blackbody temperature images and then performing calibration. Common methods include single-point correction, two-point correction, multi-point correction, and polynomial fitting correction. Infrared detectors are generally calibrated using these methods before being put into use. However, during operation, changes in external factors such as temperature and voltage can alter the detector's response coefficient. Continuing with calibration-based correction in such cases necessitates interrupting equipment use for recalibration, severely impacting the normal operation of the infrared detector. To address this issue, scene-based non-uniform correction methods utilize the texture characteristics of the captured scene pixels to perform real-time correction, eliminating the need for blackbody recalibration.

[0004] Most current non-uniform correction algorithms in the infrared field are designed for infrared external array detectors. The difficulty in removing periodic noise from infrared linear circular scan detectors lies in:

[0005] 1. An infrared perimeter-scan linear detector generates tens of thousands of images per revolution, resulting in a large data volume. This necessitates high real-time performance of the correction algorithm. Traditional methods using frequency domain filters involve computationally intensive time-frequency conversion, making them unsuitable for real-time requirements.

[0006] 2. The non-uniform noise of infrared circular scan linear detectors is in the form of horizontal stripes. If time-frequency conversion is used to process the frequency domain, it is easy to damage the texture details of the image, blur the image, and even produce ghost artifacts.

[0007] 3. Multi-frame image denoising algorithms need to consider factors such as camera shake and scene thermal radiation changes, which poses challenges to the real-time performance and robustness of the algorithms. Summary of the Invention

[0008] To address the shortcomings of existing technologies, the present invention aims to provide a non-uniformity correction method for infrared linear detectors based on guided filtering, which offers high real-time performance and effective noise reduction.

[0009] To solve the above problems, the technical solution of the present invention is as follows:

[0010] A method for correcting non-uniformity in an infrared linear detector includes the following steps:

[0011] Extract a portion of the infrared perimeter scan image from the original infrared perimeter scan image;

[0012] The selected infrared circular scan image is denoised using a guided filtering model.

[0013] The linear correction coefficients are calculated using a linear fitting model based on guided filtering.

[0014] The obtained linear correction coefficients are used to perform non-uniformity correction on the entire frame of infrared circumferential scan image.

[0015] Preferably, in the step of denoising the selected infrared circular scan image using a guided filtering model, the guided image and the output image of the guided filter have a linear relationship within a small window range, as expressed by the formula: In the formula, q i It outputs the pixel values ​​of the image, (a k ,b k ) are the linear coefficients within the small window, I i It is the guide image pixel value, w k It is the window of the pilot filter.

[0016] Preferably, in the step of denoising the selected infrared circular scan image using a guided filtering model, the coefficient (a) k ,b k The calculation formulas for ) are as follows:

[0017]

[0018]

[0019] Where: μ k and It is the guide image v in window w k The mean and variance in the data. The input image p i In window w k The mean of the window w, |w| is the mean of the window w. k The number of pixels in the window, ε is a manually set parameter that determines the degree of filtering applied to the image by the guided filter within the window.

[0020] Preferably, in the step of denoising the selected infrared circular scan image using a guided filtering model, the noisy image I... p As the guide image and input image, the images are smoothed through a one-dimensional vertical window, and the coefficients (a) are... k ,b k The calculation formula for ) becomes:

[0021]

[0022] b' k =(1-a') k )·μ k

[0023] In the formula, μ k and Image I p In window w k The mean and variance of the data.

[0024] Preferably, in the step of denoising the selected infrared circular scan image using a guided filtering model, the noisy image I... p The guide image and input image are used for guided filtering, and the output image is the vertically smoothed image I. u Using I p I u Subtraction yields image I containing non-uniform noise and some high-frequency texture. n I n =I p -I u , will I u As a guide image, I n The input image is subjected to guided filtering to obtain the output image I. s Using I p and I s Subtracting the two results in the image I after guided filtering and denoising. q I q =I p -I s .

[0025] Preferably, in the step of calculating the linear correction coefficients using a linear fitting model based on guided filtering, the linear correction coefficients a(i) and b(i) are calculated using the following formulas:

[0026]

[0027]

[0028] In the formula: W is the image width of the linear circumferential scan image after guided filtering and denoising processing, I q(i,j) is the denoised image after guided filtering, I p (i, j) is I q (i,j) corresponds to the unprocessed raw infrared circumferential scan image containing stripe noise.

[0029] Preferably, in the step of performing non-uniformity correction on the entire frame of infrared circumferential scan image using the obtained linear correction coefficients, the formula for performing non-uniformity correction on the entire frame of infrared circumferential scan image based on the obtained linear correction coefficients a(i) and b(i) is as follows:

[0030]

[0031] In the formula: a(i) and b(i) are the correction coefficients of the i-th row of the image. It is a single-frame periodic scan image containing noise and not processed. It is the final image after linear model correction.

[0032] Furthermore, the present invention also provides an infrared linear detector non-uniformity correction device, characterized in that the device includes a processor and a memory for storing executable instructions of the processor, the processor being configured to perform the infrared linear detector non-uniformity correction method as described above by executing the executable instructions.

[0033] Compared with the prior art, the present invention has the following advantages:

[0034] 1. This invention has high real-time performance. It completes the non-uniformity correction of the entire frame of the circumferential scan image by calculating the correction coefficient using only a portion of the line column image, which well meets the real-time requirements of high-resolution circumferential scan line column images with tens of thousands of columns per frame.

[0035] 2. The present invention has good noise reduction effect, and the designed linear correction model conforms to the structural characteristics of each row of imaging and fixed detection unit in the linear circumferential scan image.

[0036] 3. The present invention is robust. The method of the present invention is based on a single-frame image denoising algorithm, and does not need to consider the impact of camera shake and scene thermal radiation changes on the denoising effect in multi-frame denoising algorithms. Attached Figure Description

[0037] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0038] Figure 1 This is a flowchart of an infrared linear detector non-uniformity correction method provided in an embodiment of the present invention;

[0039] Figure 2These are partial infrared circumferential scan images provided in embodiments of the present invention;

[0040] Figure 3 This is a schematic diagram of the guided filtering process provided in an embodiment of the present invention. Detailed Implementation

[0041] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0042] Specifically, the present invention provides a method for correcting the non-uniformity of an infrared linear detector, such as... Figure 1 As shown, the method includes the following steps:

[0043] S1: Extract a portion of the infrared circumferential scan image from the original infrared circumferential scan image;

[0044] Specifically, a portion of the infrared perimeter scan image is extracted from the original infrared perimeter scan image, such as... Figure 2 As shown.

[0045] S2: Denoise the selected infrared circular scan image using a guided filtering model;

[0046] Specifically, a guided filtering model is used to denoise the selected infrared circular scan image to eliminate non-uniform horizontal stripe noise in the image.

[0047] 1. Basic knowledge of guided filtering

[0048] The basis for guided filtering is the assumption that there is a linear relationship between the guide image and the output image within a small window, as shown in the following formula:

[0049]

[0050] In the formula: q i It outputs the pixel values ​​of the image, (a k ,b k ) are the linear coefficients within the small window, I i It is the guide image pixel value, w k This is the window of the guided filter; its shape needs to be designed based on the characteristics of the noise to be removed. The output image q is constrained. i With input image p i The goal is to achieve the guided filtering by minimizing the mean square error.

[0051] coefficient (a) k ,b kThe calculation formulas for ) are shown in formulas (2) and (3) below:

[0052]

[0053]

[0054] Where: μ k and It is the guide image v in window w k The mean and variance in the data. The input image p i In window w k The mean of the window w, |w| is the mean of the window w. k The number of pixels in the window, ε is a manually set parameter that determines the degree of filtering applied to the image by the guided filter within the window.

[0055] 2. Linear detector guided filtering model

[0056] As shown in Figure 3, since the non-uniform noise in the linear circumferential scan image is horizontal stripe noise, in this embodiment, the noise image I is first... p As the guide image and input image, the images are smoothed through a one-dimensional vertical window, transforming formulas (2) and (3) into formulas (4) and (5):

[0057]

[0058] b' k =(1-a') k )·μ k (5)

[0059] μ k and Image I p In window w k The mean and variance of the image are calculated. Guided filtering is performed using equations (4) and (5), and the output image is the vertically smoothed image I. u Using I p I u Subtraction yields image I containing non-uniform noise and some high-frequency texture. n As shown in the following formula:

[0060] I n =I p -I u (6)

[0061] To better understand image I n To extract transverse non-uniform stripe noise, this invention designs a one-dimensional transverse guided filter window to extract I... u As a guide image, I nThe input image is subjected to guided filtering to obtain the output image I. s At this point, we consider image I to be... s It contains the vast majority of horizontal non-uniform stripe noise and a small amount of high-frequency horizontal texture details of the image, using I p and I s Subtracting the two results in the image I after guided filtering and denoising. q As shown in the following formula.

[0062] I q =I p -I s (7)

[0063] S3: Calculate the linear correction coefficient using a linear fitting model based on guided filtering;

[0064] Specifically, the linear correction coefficients a(i) and b(i) are calculated using a linear fitting model based on guided filtering, and the fitting coefficients are calculated using linear regression, as shown in equations (8) and (9) below:

[0065]

[0066]

[0067] In the formula: W is the image width of the linear circumferential scan image after guided filtering and denoising processing, I q (i,j) is the denoised image after guided filtering, I p (i, j) is I q (i,j) corresponds to the unprocessed raw infrared circumferential scan image containing stripe noise.

[0068] S4: Using the obtained linear correction coefficients, perform non-uniformity correction on the entire frame of infrared circumferential scan image.

[0069] Specifically, using the obtained linear correction coefficients a(i) and b(i), the non-uniformity correction is performed on the entire frame of infrared circular scan image, as shown in the following formula:

[0070]

[0071] In the formula: a(i) and b(i) are the correction coefficients of the i-th row of the image. It is a single-frame periodic scan image containing noise and not processed. It is the final image after linear model correction.

[0072] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. An infrared linear array non-uniformity correction method, comprising: The method comprises the following steps: cutting part of the infrared line-scan image from the original infrared line-scan image; using a guided filter model to denoise the selected infrared line-scan image; A linear correction coefficient is calculated using a linear fitting model based on a guided filter, the linear correction coefficient , , and the calculation formula is: ； ； In the formula, W is the image width of the line-by-line circumferential scanning image after the guided filter denoising processing, is the denoised image after the guided filter denoising processing, is The corresponding original infrared line-by-line circumferential scanning image containing the stripe noise is not processed. using the obtained linear correction coefficient to correct non-uniformity of the whole infrared line-scan image.

2. The infrared linear detector non-uniformity correction method according to claim 1, wherein, In the step of denoising the selected infrared column image by using the guided filter model, the guided image and the output image of the guided filter have a linear relationship in a small window range, and the formula is: , wherein, is an output image pixel value, is a linear coefficient in a small window, is a guided image pixel value, is a window of the guided filter.

3. The method of claim 2, wherein the non-uniformity correction is performed for each of the plurality of infrared rays. The calculation formula of the coefficient in the step of denoising the selected infrared column image by using the guided filtering model is respectively: ​ ； ； wherein: and is the guided image is the mean value and variance of the input image in the window is the mean value of the input image in the window is the mean value of the input image is the number of pixels in the window ε is a manually set parameter, which determines the degree of filtering of the image in the window by the guided filter.

4. The infrared linear array non-uniformity correction method according to claim 3, wherein, In the step of denoising the selected infrared column image by using the guided filtering model, the noise image is denoised by using the guided filtering model As the guided image and the input image, the image is smoothed by a one-dimensional longitudinal window, and the coefficient The calculation formula is changed into: ； ； wherein and is the image In the window the mean and variance.

5. The infrared linear array non-uniformity correction method of claim 4, wherein, In the step of denoising the selected infrared column image by using the guided filter model, the noise image As a guide image and an input image Performing guided filtering, the output image is a smoothed image in the vertical direction , using , Subtracting to obtain an image containing non-uniform noise and part of the image high-frequency texture , , as a guide image, As an input image, Performing guided filtering to obtain an output image , using And Subtracting to obtain a guided filter denoised image , , .

6. The method of claim 1, wherein, The linear correction coefficient obtained is used in the step of correcting the non-uniformity of the whole-frame infrared line-by-line scanning image, and the linear correction coefficient is used in the formula for correcting the non-uniformity of the whole-frame infrared line-by-line scanning image 、 The formula for correcting the non-uniformity of the whole-frame infrared line-by-line scanning image is: ； wherein: , is the correction coefficient of the i-th row of the image, is the single frame weekly scanning image containing noise without processing, is the final image after linear model correction.

7. A non-uniformity correction device for an infrared linear detector, characterized in that, The device comprises a processor and a memory for storing executable instructions of the processor, and the processor is configured to execute the infrared line-scan non-uniformity correction method as claimed in any one of claims 1 to 6 by executing the executable instructions.

Images