An Infrared Image Non-Uniformity Correction Method Based on Image Gray Gradient

Abstract

The present invention discloses an infrared image non-uniformity correction method based on image gray gradient, which includes the following steps: acquiring two frames of images, performing two-point non-uniformity correction on the two frames of images; extracting the image gray gradient features of the corrected images by using a gradient operator; finding the brightest point in the feature map, expanding N pixels up, down, left, and right centered on the brightest point to obtain an image block of a specified size, and performing a binarization operation on the image block; performing row and column projection matching on the two frames of binarized images to obtain offset information; according to the offset information, subtracting the overlapping parts of the two frames of images, calculating the compensation amount required to update the background, and correcting the background; using the corrected background to perform non-uniformity correction on the current frame of image. The beneficial effects of the present invention are: effectively eliminating fixed noise, high-frequency noise, and low-frequency noise, improving the efficiency and accuracy of non-uniformity correction, saving computing resources, and not requiring a shutter as a reference.

Description

Technical Field

[0001] The present invention relates to the field of infrared image correction, and particularly to an infrared image non-uniformity correction method based on image gray gradient. Background Art

[0002] Infrared focal plane detectors are the core components of infrared thermal imaging systems and have extensive applications in the military and civilian fields. However, in the actual application process, the original output images of infrared thermal imagers have obvious non-uniformity, and the main reasons for its generation are the non-uniformity of the detector itself, amplifier noise, and background radiation photon noise. How to effectively reduce the non-uniformity of images in the infrared thermal imaging process and improve the image quality is a problem that needs to be solved by those skilled in the art. Currently, a commonly used method for non-uniformity correction (Non-uniformity correction, abbreviated as NUC) is the two-point calibration method, which performs a two-point calibration before the infrared thermal imager leaves the factory. However, during the use process, with the changes in the environment and the detector itself, there are differences between the non-uniformity of the image and the calibration result. Therefore, it is necessary to perform shutter correction regularly, using the shutter as the standard, and frequently perform calibration for non-uniformity correction, but this will increase the power consumption and cost of the system. Summary of the Invention

[0003] In order to solve the above problems, the present invention provides an infrared image non-uniformity correction method based on image gray gradient. This method collects two frames of images, performs two-point NUC correction on the two frames of images to reduce the influence of non-uniformity residues in the relatively poor background on subsequent algorithms, uses the Prewitt operator to extract the image gray gradient features, finds the brightest point in the feature map, and expands 128 pixels up, down, left, and right centered on the brightest point to obtain an image block of 256×256 size, and performs binarization operations. After performing the binarization operations, the previous frame and the current frame are respectively subjected to row and column projection matching to calculate the displacement to obtain the offset information; according to the offset information, the overlapping parts of the two frames of images are subtracted, the compensation amount required for updating the background is calculated, the background is updated, and the current frame is subjected to NUC correction using the corrected background. This can effectively eliminate fixed noise, high-frequency noise, and low-frequency noise, improve the efficiency and accuracy of non-uniformity correction, eliminate the need for a shutter, and save computing resources.

[0004] An infrared image non-uniformity correction method based on image gray gradient provided by the present application includes the following steps:

[0005] S1. Collect two frames of images, perform two-point non-uniformity NUC correction on the two frames of images to obtain two corrected frames of images;

[0006] S2. Use the Prewitt operator to extract the image gray gradient features of the corrected images to obtain a feature map;

[0007] S3. Locate the brightest point in the feature map, expand N pixels up, down, left, and right centered on the brightest point to obtain an image block of size 2N×2N, and perform a binarization operation on the image block to obtain two frames of binarized images;

[0008] S4: Perform row and column projection matching on the two frames of binarized images to obtain offset information;

[0009] S5: According to the offset information, subtract the overlapping parts of the two frames of images, calculate the compensation amount required to update the background, and correct the background;

[0010] S6: Use the corrected background to perform non-uniform NUC correction on the current frame image to obtain the non-uniform NUC corrected image of the current frame.

[0011] Further, in step S1, the formula for performing two-point non-uniform NUC correction on the two frames of images is shown in Equation (1):

[0012] PicNUC n =(pic n -base)*K = pic n *K - base*K (1)

[0013] where the subscript n represents the serial numbers of the previous frame and the current frame, n = 1, 2; PicNUC n is the image after NUC correction; pic n is the image before correction; K is the pre-stored gain coefficient, and base is the background image read in the actual situation.

[0014] Further, the formula for obtaining the feature map in step S2 is:

[0015]

[0016]

[0017]

[0018] where fx and fy are the gray gradient feature maps in the horizontal and vertical directions respectively, A is the input image, and f is the final gray gradient feature map.

[0019] Further, in step S3, the formula for obtaining the binarized image is:

[0020]

[0021] Among them, M is an image block of 2N×2N. If the gray level of the input image is ≥Y max, it is set to 1, otherwise it is 0; max(M) represents the maximum gray value in the image block; mean(M) represents the average gray value in the image block.

[0022] Further, step S4 is specifically as follows:

[0023] Perform row and column projection matching on the previous frame's initial template and the current frame's search area after binarization operation. The displacement calculation range is [-10, +10]. The displacement amount is based on the previous frame, moving the current frame, with the length remaining the same as that of the previous frame. Calculate the sum of the absolute values of the differences one by one to obtain the minimum error value. Compare the obtained minimum error value with the set threshold sadlimit. If the actually obtained best position error value is greater than this threshold, output the displacement amount as 0 and mark it; otherwise, the position corresponding to this error value is the required displacement offset, that is, the global inter-frame displacement (x, y) of the image, where x and y represent the displacement amounts of the entire current frame compared to the entire previous frame in the horizontal and vertical directions.

[0024] Further, in step S5, the calculation formula for correcting the background compensation amount is as follows:

[0025] picK1 = pic1 * K

[0026] picK2 = pic2 * K

[0027] baseNow = base * K

[0028] RawModify = [picK1(i, j) - picK2(i + x, j + y)] - [baseNow(i, j) - baseNow(i + x, j + y)]

[0029] Among them, pic1 and pic2 are the original 14-bit images of the previous frame and the current frame respectively; RawModify is the compensation amount that the background needs to be corrected; base is the background image.

[0030] Further, the background compensation amount is restricted by the update rule, specifically as follows:

[0031] If |RawModify| ≥ 2×ModLimit, where ModLimit is the preset background update threshold, then the background compensation amount is updated to RawModify * :

[0032]

[0033] If |RawModify| < 2×ModLimit, then the background compensation amount is updated to RawModify* :

[0034]

[0035] According to the updated background compensation amount, the finally obtained compensation amount matrix is:

[0036]

[0037] Where Modify is initialized as a zero matrix of the corresponding size, and the finally updated background BaseN is: BaseN = baseNow + Modify.

[0038] In step S6, the calculation formula for obtaining the non-uniform NUC correction image of the current frame is as follows:

[0039] PicNuc = pic * K - BaseN

[0040] Where PicNuc is the non-uniform NUC correction image of the current frame.

[0041] Compared with the prior art, the beneficial effects of the present invention include: effectively eliminating fixed noise, high-frequency noise and low-frequency noise, improving the efficiency and accuracy of non-uniformity correction, saving computing resources, and eliminating the need for a shutter as a reference. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0044] Please refer to Figure 1 , Figure 1 which is a flowchart of the method of the present invention; the present invention includes the following steps:

[0045] S1. Collect two frames of images, perform two-point non-uniformity NUC correction on the two frames of images, and obtain two frames of corrected images;

[0046] It should be noted that in step S1, collecting two frames of images and performing two-point NUC correction on the two frames of images reduces the non-uniformity residue in the poor background and affects the subsequent algorithm. Collecting two frames of images and performing two-point NUC correction on the two frames of images reduces the non-uniformity residue in the poor background and affects the subsequent algorithm; the formula for performing two-point non-uniform NUC correction on the two frames of images is shown in Equation (1):

[0047] PicNUCn = (pic n -base) * K = pic n * K - base * K (1)

[0048] Where the subscript n represents the previous frame and the current frame numbers, n = 1, 2; PicNUC n is the image after NUC correction; pic n is the image before correction; K is the pre-stored gain coefficient, and base is the background image read under actual conditions.

[0049] S2. Use the Prewitt operator to extract the image gray gradient features of the corrected image to obtain a feature map;

[0050] It should be noted that in step S2, the Prewitt operator is used to extract the image gray gradient features of the two collected frames of images, which is realized by performing convolution and operation with each pixel point in the image through a 3×3 convolution template in the horizontal and vertical directions; the calculation formula for obtaining the feature map is:

[0051]

[0052]

[0053]

[0054] Where fx and fy are the gray gradient feature maps in the horizontal and vertical directions respectively, A is the input image, and f is the final gray gradient feature map.

[0055] S3. Find the brightest point of the feature map, and expand N pixels up, down, left, and right centered on the brightest point to obtain an image block of size 2N×2N, and perform a binarization operation on the image block to obtain two frames of binarized images;

[0056] As an embodiment, in the present invention, N = 128;

[0057] In step S3, find the maximum value of the feature map, and expand 64 pixels up, down, left, and right centered on the maximum value position to obtain an image block of size 128×128, and perform a binarization operation. The binarization gray threshold Ymax is 1 / 2 of the sum of the maximum gray value and the gray mean value in the image block to achieve binarization. Its formula is shown in formula (3):

[0058]

[0059] Where M is an image block of 128×128, pixels with input image gray level >= Ymax are set to 1, otherwise 0.

[0060] S4: Perform row and column projection matching on two frames of binary images to obtain offset information;

[0061] It should be noted that step S4 is specifically as follows:

[0062] Perform row and column projection matching on the previous frame of the initial template and the current frame of the search area after binarization operation respectively. The displacement calculation range is [-10, +10]. The displacement is based on the previous frame, moving the current frame, with the length consistent with that of the previous frame. Calculate the sum of the absolute values of the differences one by one to obtain the minimum error value, and compare the obtained minimum error value with the set threshold sadlimit. If the actual best position error value obtained is greater than this threshold, output the displacement as 0 and mark it; otherwise, the position corresponding to this error value is the required displacement offset, that is, the global image inter-frame displacement (x, y), where x and y represent the displacement amounts of the entire current frame image compared to the entire previous frame image in the horizontal and vertical directions.

[0063] S5: According to the offset information, perform subtraction operation on the overlapping parts of the two frames of images, calculate the compensation amount required to update the background, and correct the background;

[0064] It should be noted that according to the offset information, perform subtraction operation on the overlapping parts of the two frames of images, calculate the compensation amount required to update the background, and update the background. Specifically: x and y represent the displacement amounts of the entire current frame image compared to the entire previous frame image in the horizontal and vertical directions.

[0065] Then the calculation of the compensation amount required to update the background is as follows:

[0066] picK1 = pic1 * K

[0067] picK2 = pic2 * K

[0068] baseNow = base * K

[0069] RawModify = [picK1(i, j) - picK2(i + x, j + y)] - [baseNow(i, j) - baseNow(i + x, j + y)]

[0070] Among them, pic1 and pic2 are the original 14-bit images of the previous frame and the current frame respectively; RawModify is the compensation amount required to correct the background; base is the background image.

[0071] To avoid calculation errors during target matching, resulting in the rapid deterioration of the background, limit its update value. ModLimit is the background update threshold, and the update rule is as follows:

[0072] If |RawModify| ≥ 2 × ModLimit, where ModLimit is the preset background update threshold, then the background compensation amount is updated to RawModify * :

[0073]

[0074] If |RawModify| < 2 × ModLimit, then the background compensation amount is updated to RawModify * :

[0075]

[0076] According to the updated background compensation amount, the finally obtained compensation amount matrix is:

[0077]

[0078] Where Modify is initialized as a zero matrix of the corresponding size, and the finally updated background BaseN is: BaseN = baseNow + Modify.

[0079] S6: Use the corrected background to perform non-uniform NUC correction on the current frame image to obtain the current frame non-uniform NUC corrected image.

[0080] In step S6, the calculation formula for obtaining the current frame non-uniform NUC corrected image is as follows:

[0081] PicNuc = pic * K - BaseN

[0082] Where PicNuc is the current frame non-uniform NUC corrected image.

[0083] The beneficial effects of the present invention are: effectively eliminating fixed noise, high-frequency noise and low-frequency noise, improving the efficiency and accuracy of non-uniformity correction, saving computing resources, and not requiring a shutter as a reference.

[0084] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0085] The above specific implementation manners of the present invention do not constitute a limitation to the protection scope of the present invention. Any other corresponding changes and deformations made according to the technical concept of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. An infrared image non-uniformity correction method based on image gray gradient, characterized in that: It includes the following steps: S1. Collect two frames of images, perform two-point non-uniformity NUC correction on the two frames of images to obtain two frames of corrected images; S2. Use the Prewitt operator to extract the image gray gradient features of the corrected images to obtain a feature map; S3. Find the brightest point in the feature map, expand N pixels up, down, left, and right centered on the brightest point to obtain an image block of size 2N×2N, and perform a binarization operation on the image block to obtain two frames of binarized images; S4: Perform row and column projection matching on the two frames of binarized images to obtain offset information; S5: According to the offset information, subtract the overlapping parts of the two frames of images, calculate the compensation amount required to update the background, and correct the background; S6: Use the corrected background to perform non-uniform NUC correction on the current frame image to obtain the non-uniform NUC corrected image of the current frame.

2. The non-uniformity correction method for infrared images based on image gray gradient according to claim 1, characterized in that: In step S1, the formula for performing two-point non-uniform NUC correction on the two frames of images is shown in formula (1): PicNUC n =(pic n -base)*K = pic n *K - base*K(1) Among them, the subscript n represents the previous frame and the current frame numbers, n = 1, 2; PicNUC n is the image after NUC correction; pic n is the image before correction; K is the pre-stored gain coefficient, and base is the background image read under actual conditions.

3. The non-uniformity correction method for infrared images based on image gray gradient according to claim 1, wherein: The calculation formula for obtaining the feature map in step S2 is: where fx and fy are the gray gradient feature maps in the horizontal and vertical directions respectively, A is the input image, and f is the final gray gradient feature map.

4. The non-uniformity correction method for infrared images based on image gray gradient according to claim 1, wherein: In step S3, the formula for obtaining the binarized image is: where M is an image block of 2N×2N, set to 1 if the input image gray level >= Y max, otherwise 0; max(M) represents the maximum gray value in the image block; mean(M) represents the average gray value in the image block.

5. The non-uniformity correction method for infrared images based on image gray gradient according to claim 1, wherein: Step S4 is specifically: Perform row and column projection matching on the initial template of the previous frame and the search area of the current frame after the binarization operation respectively, where the displacement calculation range is [-10, +10], the displacement amount is based on the previous frame, move the current frame, and the length is the same as the length of the previous frame. Calculate the sum of the differences and absolute values one by one to obtain the minimum error value, and compare the obtained minimum error value with the set threshold sadlimit. If the actual obtained best position error value is greater than the threshold, output the displacement amount as 0 and mark it; Otherwise, the position corresponding to the error value is the required displacement offset, that is, the global image inter-frame displacement (x, y), where x and y represent the displacement amounts of the entire current frame image in the horizontal and vertical directions compared to the entire previous frame image.

6. The non-uniformity correction method for infrared images based on image gray gradient according to claim 5, characterized in that: In step S5, the calculation formula for correcting the background compensation amount is as follows: picK1 = pic1 * K picK2 = pic2 * K baseNow = base * K RawModify = [picK1(i, j) - picK2(i + x, j + y)] - [baseNow(i, j) - baseNow(i + x, j + y)] where pic1 and pic2 are the original 14-bit images of the previous frame and the current frame respectively; RawModify is the compensation amount that the background needs to be corrected; base is the background image.

7. The non-uniformity correction method for infrared images based on image gray gradient according to claim 6, wherein: The background compensation amount is restricted by the update rule, specifically: If |RawModify| ≥ 2 × ModLimit, where ModLimit is the preset background update threshold, then the background compensation amount is updated to RawModify * : If |RawModify| < 2 × ModLimit, the background compensation amount is updated to RawModify * : According to the updated background compensation amount, the final obtained compensation amount matrix is: Among them, Modify is initialized as a zero matrix of the corresponding size, and the finally updated background BaseN is: BaseN = baseNow + Modify*.

8. The non-uniformity correction method for infrared images based on image gray scale gradient according to claim 7, characterized in that: In step S6, the calculation formula for obtaining the non-uniform NUC correction image of the current frame is as follows: PicNuc = pic * K - BaseN, where PicNuc is the non-uniform NUC correction image of the current frame.

Images