Image gain non-uniformity correction method and device for staring infrared detector

Abstract

The present invention provides a method and device for correcting image gain non-uniformity of a registration-type staring infrared detector, which can improve the accuracy of non-uniformity correction of infrared images. The method includes: constructing a longitudinal gain difference matrix through the longitudinal difference of infrared detector signals output by the same scene radiation in the previous and next frame images, further transforming and calculating the longitudinal gain difference matrix to obtain an average longitudinal gain correction matrix, and performing longitudinal correction on the image through the average longitudinal gain correction matrix, constructing a transverse gain difference matrix based on the transverse difference of infrared detector signals output by the same scene radiation of the longitudinally corrected image, and transforming the transverse gain difference matrix to obtain an average transverse gain correction matrix, determining an average gain correction matrix according to the average longitudinal gain correction matrix and the average transverse gain correction matrix, and performing gain non-uniformity correction on the input image through the average gain correction matrix.

Description

Technical Field

[0001] The present invention relates to the field of infrared image processing technology, and in particular to a method and device for correcting image gain non-uniformity of a staring type infrared detector. Background Art

[0002] The staring infrared imaging detector is composed of a detector focal plane array. Due to the limitations of the manufacturing process, each detector in the infrared focal plane has a corresponding problem of non-uniformity. The non-uniform response of the staring infrared detector is ultimately manifested as a dot-shaped fixed pattern noise on the infrared image. In order to overcome this defect, calibration technology is widely used in the correction of non-uniformity. This technology guides the infrared camera to image black bodies of various temperature levels to ensure a uniform temperature distribution within the field of view of the camera. On this basis, a linear or high-order fitting algorithm is used to calculate the gain and bias parameters of each detector to achieve the purpose of compensating for the non-uniformity. It is worth noting that the spatial non-uniform response of the focal plane will drift slowly over time, so the correction parameters must be continuously estimated and updated, otherwise the fixed pattern noise will appear again in the output image frame. If the calibration correction technology is used to update the correction parameters, the field of view will be regularly blocked by the black body target during the use of the infrared camera, resulting in the loss of key scene information during calibration, which cannot fully meet the requirements of information integrity in practical applications.

[0003] In order to circumvent the limitations of blackbody calibration technology, scene-based non-uniformity correction methods use scene images captured by infrared detectors to dynamically estimate and remove non-uniformity. In scene-based non-uniformity correction algorithms, traditional methods can be divided into three main categories: First, methods based on constant statistics assume that the time domain mean and variance of all detectors are identical. To meet this assumption, the captured infrared scene information needs to be sufficiently diverse, which is often difficult to fully meet in practical applications, resulting in deviations in correction parameters. Second, neural network-based methods build a neural network model and use scene information to train the correction parameters of the model, but this method has the problems of high computing power requirements and poor convergence; third, image registration-based methods use inter-frame motion information to align moving scene images and determine the correction coefficient based on the signal response difference between adjacent frames. However, this type of algorithm can usually only correct bias non-uniformity and ignores the gain non-uniformity problem remaining in the corrected image. Therefore, how to more effectively use scene information to optimize the non-uniformity correction of infrared images is still a technical problem to be solved. Summary of the invention

[0004] The embodiment of the present invention provides a method and device for correcting image gain non-uniformity of a staring infrared detector, which can improve the accuracy of image non-uniformity correction.

[0005] A first aspect of the present invention provides a method for correcting image gain non-uniformity of a staring infrared detector, comprising:

[0006] Determine a first detector output signal of a first frame image after correcting bias non-uniformity, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any two adjacent infrared frames of images having longitudinal sub-pixel translation in the multiple frames after correcting bias non-uniformity;

[0007] Determine the output signal of the first registration detector of the second frame image according to the output signal of the second detector and the inter-frame sub-pixel longitudinal motion translation;

[0008] Constructing a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector and the longitudinal motion translation of the sub-pixel between frames;

[0009] Transforming and column-wise multiplying the longitudinal gain difference matrix to obtain a longitudinal gain correction matrix;

[0010] Adding the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number;

[0011] Calculating an average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix;

[0012] Performing longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared images with lateral sub-pixel shift in the multi-frame image after the bias non-uniformity is corrected;

[0013] Determine the second registration detector output signal of the fourth frame image according to the fourth detector output signal and the inter-frame sub-pixel lateral motion translation

[0014] Constructing a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector and the inter-frame sub-pixel lateral motion translation;

[0015] Transforming and row-wise multiplying the lateral gain difference matrix to obtain a lateral gain correction matrix;

[0016] Adding the lateral gain correction matrix to a second first-in-first-out queue until the number of lateral gain correction matrices stored in the second first-in-first-out queue reaches a second preset number;

[0017] Calculating an average matrix of the second preset number of lateral gain correction matrices in the second FIFO queue to obtain an average lateral gain correction matrix;

[0018] Determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix;

[0019] The gain non-uniformity correction is performed on the input image according to the average gain correction matrix, and the input image is an infrared image after the bias non-uniformity correction is performed.

[0020] A second aspect of the present invention provides an image gain non-uniformity correction device for a staring type infrared detector, comprising:

[0021] A determination module, used to determine a first detector output signal of a first frame image after bias non-uniformity correction, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any two adjacent infrared frames of images having longitudinal sub-pixel translation in a plurality of frames after bias non-uniformity correction;

[0022] The determination module is further used to determine the output signal of the first registration detector of the second frame image according to the output signal of the second detector and the inter-frame sub-pixel longitudinal motion translation;

[0023] A matrix construction module, used for constructing a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector and the longitudinal motion translation of the sub-pixel between frames;

[0024] A transformation and multiplication module, used for transforming and multiplying the longitudinal gain difference matrix by columns to obtain a longitudinal gain correction matrix;

[0025] An adding module, configured to add the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number;

[0026] A calculation module, used for calculating an average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix;

[0027] a correction module, used for performing longitudinal correction on a third frame image and a fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared frames of images having lateral sub-pixel shift in the multi-frame images after the bias non-uniformity is corrected;

[0028] The determination module is further used to determine the output signal of the second registration detector of the fourth frame image according to the output signal of the fourth detector and the inter-frame sub-pixel lateral motion translation amount.

[0029] The matrix construction module is further used to construct a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector and the inter-frame sub-pixel lateral motion translation;

[0030] The transformation and multiplication module is further used to transform and multiply the lateral gain difference matrix row by row to obtain a lateral gain correction matrix;

[0031] The adding module is further used to add the lateral gain correction matrix to a second first-in-first-out queue until the lateral gain correction matrix stored in the second first-in-first-out queue reaches a second preset number;

[0032] The calculation module is further used to calculate the average matrix of the second preset number of lateral gain correction matrices in the second FIFO queue to obtain an average lateral gain correction matrix;

[0033] The determination module is further configured to determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix;

[0034] The correction module is further used to perform gain non-uniformity correction on the input image according to the average gain correction matrix, and the input image is an infrared image after the bias non-uniformity is corrected.

[0035] The third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the processor is used to implement the steps of the image gain non-uniformity correction method of a staring type infrared detector as described in the first aspect when executing a computer management program stored in the memory.

[0036] A fourth aspect of the present invention provides a computer-readable storage medium having a computer management program stored thereon, which, when executed by a processor, implements the steps of the image gain non-uniformity correction method for a staring type infrared detector as described in the first aspect above.

[0037] In summary, it can be seen that in the embodiment provided by the present application, a longitudinal gain difference matrix is ​​constructed by the longitudinal difference of the infrared detector signals output by the same scene radiation in the previous and next frame images, and an average longitudinal gain correction matrix is ​​obtained by further transforming and calculating the longitudinal gain difference matrix, and the image is longitudinally corrected by the average longitudinal gain correction matrix, and a lateral gain difference matrix is ​​constructed based on the lateral difference of the infrared detector signals output by the same scene radiation in the image after the longitudinal correction, and an average lateral gain correction matrix is ​​obtained by further transforming and calculating the lateral gain difference matrix, and finally the average gain correction matrix is ​​determined according to the average longitudinal gain correction matrix and the average lateral gain correction matrix, and finally the input image is corrected for gain non-uniformity by the average gain correction matrix, so as to solve the problem of residual gain non-uniformity in the image output by the traditional infrared non-uniformity correction method based on registration, further correct the gain non-uniformity, and improve the accuracy of infrared image non-uniformity correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A schematic diagram of a flow chart of a method for correcting image gain non-uniformity of a staring infrared detector provided by an embodiment of the present invention;

[0039] Figure 2 A virtual structural diagram of an image gain non-uniformity correction device for a staring type infrared detector provided by an embodiment of the present invention;

[0040] Figure 3 A schematic diagram of the hardware structure of the image gain non-uniformity correction device for a staring type infrared detector provided by an embodiment of the present invention;

[0041] Figure 4 A schematic diagram of an electronic device according to an embodiment of the present invention;

[0042] Figure 5 A schematic diagram of an embodiment of a computer-readable storage medium provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0044] The terms "first", "second", and "third" in the present invention are used to distinguish different objects rather than to describe a specific order. In addition, the terms "include", "have", and any variations thereof are intended to cover non-exclusive inclusions.

[0045] The image gain non-uniformity correction method of the staring type infrared detector provided by the present invention is explained below from the perspective of the image gain non-uniformity correction device of the staring type infrared detector. The image gain non-uniformity correction device of the staring type infrared detector can be a server or a service unit in the server, and there is no specific limitation.

[0046] See also Figure 1 , Figure 1 A schematic flow chart of a method for correcting image gain non-uniformity of a staring infrared detector provided in an embodiment of the present invention, wherein the method for correcting image gain non-uniformity of a staring infrared detector comprises:

[0047] 101. Determine a first detector output signal of a first frame image after bias non-uniformity correction, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames.

[0048] In this embodiment, the image gain non-uniformity correction device of the staring infrared detector can determine the first detector output signal of the first frame image after the bias non-uniformity is corrected, the second detector output signal of the second frame image, and the sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any adjacent two frames of infrared images in the multi-frame infrared images with longitudinal sub-pixel translation obtained after the bias non-uniformity is corrected by the staring infrared detector non-uniformity correction method based on registration. Specifically, the image gain non-uniformity correction device of the staring infrared detector can determine the first detector output signal and the second detector output signal by the following formula:

[0049]

[0050]

[0051] in, is the output signal of the first detector, i = 1, 2, ..., M, j = 1, 2, ..., N, M × N is the resolution of the first frame image, is the real infrared radiation value of the scene corresponding to the first frame image, g(i,j) is the gain non-uniformity parameter, is the output signal of the second detector, is the actual infrared radiation value of the scene corresponding to the second frame image;

[0052] It should be noted that after correcting the bias non-uniformity, the previous and next frame images with longitudinal sub-pixel shift are and Their inter-frame sub-pixel longitudinal motion translation is (α v ,β v).

[0053] In a possible embodiment, the inter-frame sub-pixel longitudinal motion translation is required to meet relatively strict longitudinal sub-pixel translation and lateral sub-pixel translation, that is, the longitudinal translation satisfies 0<|α v |≤1; lateral translation β v is approximately equal to 0, satisfying |β v |<T mov , where T mov is the motion detection threshold. It is usually T according to the selection of the inter-frame registration algorithm and the size of the non-uniform fixed pattern noise. mov The value range of is 0.05 to 0.1. For the continuous frames that do not meet the translation requirements, they can be discarded. For the convenience of description, the subsequent method steps assume that the longitudinal translation satisfies 0<α v ≤1, that is, the scene is translated downward sub-pixel. For -1≤α v The case of <0 (the scene is translated upward sub-pixel) is similar and will not be described in detail.

[0054] The previous and next frame images with lateral sub-pixel shift after correcting the bias non-uniformity are: and Their inter-frame motion translation is (α h ,β h ).

[0055] In a possible embodiment, the image frame translation is required to satisfy a relatively strict lateral sub-pixel translation, that is, the lateral translation amount satisfies 0<|β h |≤1; longitudinal translation α h is approximately equal to 0, satisfying α h |<T mov For the continuous frames that do not meet the translation requirements, they can be discarded. For the convenience of description, the subsequent method steps assume that the lateral translation satisfies 0<β h ≤1, the scene is shifted to the right subpixel. For -1≤β h The case of <0 (the scene is shifted to the left sub-pixel) is similar and will not be described in detail.

[0056] 102. Determine a first registration detector output signal of a second frame of image according to the second detector output signal and an inter-frame sub-pixel longitudinal translation amount.

[0057] In this embodiment, after determining the second detector output signal and the inter-frame sub-pixel longitudinal running translation, the image gain non-uniformity correction device of the staring infrared detector can determine the first registration detector output signal according to the second detector output signal and the inter-frame sub-pixel longitudinal running translation. Since the infrared imaging detector has a high output frame rate and the time interval between two consecutive frames is short, the change in the radiation value of each object in the scene of the image can be ignored. Then, the scene content at the coordinate (i, j) in the kth frame and the coordinate (i-α) in the k-1th frame are v ,j-β v ) have the same scene content and the same radiation value, that is, Due to the lateral translation β v is approximately equal to 0, so the radiation relationship between the previous and next frame scenes can be simplified to

[0058] It should be noted that since the radiation value in the scene cannot be measured directly, only the detector output can be obtained, that is, where g * (i-α v ,j) is the ideal gain coefficient after longitudinal translation.

[0059] In one embodiment, a linear interpolation model is used to calculate g * (i-α v ,j) is approximately expressed as follows:

[0060] g * (i-α v ,j)=(1-α v )g(i,j)+α v g(i-1,j);

[0061] Then the output signal of the first registration detector can be expressed by the following formula:

[0062]

[0063] In one possible embodiment, Using a linear interpolation model for approximate representation, the output signal of the registration detector can be expressed by the following formula:

[0064]

[0065] 103. Construct a longitudinal gain difference matrix according to the first detector output signal, the first registration detector output signal, and the inter-frame sub-pixel longitudinal motion translation.

[0066] In this embodiment, the image gain non-uniformity correction device of the staring type infrared detector determines the output signal of the first detector. The first registration detector output signal And the sub-pixel longitudinal motion translation between frames (α v ,β v ) After that, the elements in the gain difference matrix can be calculated by the following formula, that is, the gain difference matrix D G Values ​​within the valid numerical coordinate range in:

[0067]

[0068] Among them, D v (i,j) is the longitudinal gain difference matrix D v The value at coordinate (i,j), is the output signal of the first detector, is the output signal of the first registration detector, (α v ,β v ) is the sub-pixel longitudinal motion translation between frames;

[0069] The longitudinal gain difference matrix D is calculated v After the values ​​of each element in the D v The regions that satisfy the above differential relationship among the elements in the previous and next frames are and The scene overlap area, that is, i = 2, 3, L, M, j = 1, 2, ..., N, for i = 1, let D v (1,j)=1, then D v The values ​​satisfy the following relationship

[0070]

[0071] Then we get the longitudinal gain difference matrix D v .

[0072] 104. Transform the longitudinal gain difference matrix and perform column-wise multiplication to obtain a longitudinal gain correction matrix.

[0073] In this embodiment, the goal of the longitudinal gain correction matrix is ​​to transform the longitudinal gain difference matrix so that the value of each element in the transformed matrix is ​​the value of the first frame image Y k The gain non-uniformity correction of the pixels in the corresponding row in D is the gain non-uniformity coefficient of a certain s-th row. Specifically, the image gain non-uniformity correction device of the staring infrared detector can be used for D v Perform the line break transformation in the elementary transformation and change D v The first row is swapped with the next row one by one until the original first row elements are moved to the sth row position, and the first s-1 rows of elements after the move are set to the inverse of the original values ​​to obtain the vertical transformation matrix

[0074]

[0075] Then the vertical transformation matrix Perform column-wise multiplication to obtain the longitudinal gain correction matrix The value of each element in Calculate using the following formula:

[0076]

[0077]

[0078] but The elements in it are the same as the first frame image Multiply the pixel values ​​of the corresponding rows in The first frame image can be The gain non-uniformity coefficients of all pixels in are corrected to the non-uniformity coefficients g(s,j) of the sth row.

[0079] 105. Add the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number.

[0080] In this embodiment, the image gain non-uniformity correction device of the staring type infrared detector can establish a length L v (That is, the first preset number is L v ) is a first-in-first-out queue, denoted as vFIFO (i.e., first first-in-first-out queue), which is used to store the latest L v The image gain non-uniformity correction device of the staring infrared detector queues the determined longitudinal gain correction matrix into the vFIFO until the length of the longitudinal gain correction matrix in the queue reaches L v So far, the queue stores multiple longitudinal gain correction matrices in the order of entry.

[0081] 106. Calculate an average matrix of a first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix.

[0082] In this embodiment, the image gain non-uniformity correction device of the staring infrared detector can calculate the average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue, and specifically determine the value of each element in the average longitudinal gain correction matrix by the following formula:

[0083]

[0084] in, is the value of each element in the average longitudinal gain correction matrix, is the first preset number L v The average value of different longitudinal gain correction coefficients,

[0085] After the values ​​of the elements in the average longitudinal gain correction matrix are calculated, the average longitudinal gain correction matrix can be constructed according to the values ​​of the elements.

[0086] It should be noted that whenever a new average longitudinal gain correction matrix is ​​calculated After that, clear all queue elements in vFIFO and wait for L v A new longitudinal gain correction matrix is ​​added to the first FIFO queue until the number of longitudinal gain correction matrices in the first FIFO queue reaches L again. v , it is determined whether the target average lateral gain correction matrix is ​​calculated, and the target average lateral gain correction matrix is ​​the same as the target average longitudinal gain correction matrix Correspondingly, the target average longitudinal gain correction matrix is the number of L that passed through the first FIFO queue in the previous cycle v The average longitudinal gain correction matrix is ​​calculated by the longitudinal gain correction matrix. Here, the number of longitudinal gain correction matrices in the first FIFO queue reaches L. v And clear as a cycle, if the target average lateral gain correction matrix is ​​calculated, then wait for the number of longitudinal gain correction matrices in the first FIFO queue to reach L v , calculate the average longitudinal gain correction matrix, clear the longitudinal gain correction matrix in the first FIFO queue, and wait for the corresponding average lateral gain correction matrix to be obtained, and then wait again for the number of longitudinal gain correction matrices in the first FIFO queue to reach L v , and repeat this process.

[0087] 107. Perform longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image.

[0088] In this embodiment, the image gain non-uniformity correction device of the staring infrared detector can perform longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared images with lateral sub-pixel translation in the multi-frame image after the bias non-uniformity is corrected. Specifically, the third detector output signal is determined by the following formula:

[0089]

[0090] in, is the output signal of the third detector, is the first preset number L v The average value of different longitudinal gain correction coefficients, is the actual infrared radiation value of the scene corresponding to the third frame image;

[0091] The output signal of the fourth detector is determined by the following formula:

[0092]

[0093] in, is the fourth detector output signal (the detector output signal is here in pixel value), is the first preset number L v The average value of different longitudinal gain correction coefficients, is the actual infrared radiation value of the scene corresponding to the fourth frame image.

[0094] It should be noted that the image obtained after longitudinal correction The pixels in the same column have the same gain non-uniformity. The third frame image and the fourth frame image also have this property, and are the preceding and following frame images with lateral sub-pixel shift after longitudinal correction.

[0095] 108. Construct a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector, and the inter-frame sub-pixel lateral motion translation.

[0096] In this embodiment, the lateral gain difference matrix is ​​based on the previous and next frame images with lateral sub-pixel translation after longitudinal correction. and The difference between the output signals of the same radiation scene in the previous and next frames is constructed. Since the infrared imaging detector has a high output frame rate and the time interval between two consecutive frames is short, the change in the radiation value of each object in the image scene can be ignored. The scene content at the coordinate (i, j) in the kth frame is different from the scene content at the coordinate (i-α) in the k-1th frame. h ,j-β h ) have the same scene content and the same radiation value, that is, Due to the longitudinal translation α h is approximately equal to 0, so the radiation relationship between the previous and next frame scenes can be simplified to

[0097] It should be noted that since the radiation value in the scene cannot be measured directly, only the detector output can be obtained, that is, where g* (i,j-β h ) is the ideal gain coefficient after lateral translation.

[0098] In a possible embodiment, a linear interpolation model is used to calculate g * (i,j-β h ) is approximated, and the specific representation is as follows:

[0099]

[0100] Therefore, the output signal of the first registration detector can be expressed by the following formula:

[0101]

[0102] After determining the third detector output signal, the second registration detector output signal, and the inter-frame sub-pixel lateral motion translation, the lateral gain difference matrix D can be calculated as follows: h Values ​​within the valid numerical coordinate range in:

[0103]

[0104] Among them, D h (i,j) is D h The value at coordinate (i, j), in D h In the construction formula of h ,β h )and is a known quantity.

[0105] In one possible embodiment, The linear interpolation model is used for approximate representation, and the specific expression is as follows:

[0106]

[0107] From this we can calculate D v The value of each element in .

[0108] It should be noted that D h The area that satisfies the above differential relationship is the previous and next frames and The scene overlap area, that is, i = 1, 2, ..., M, j = 2, 3, ..., N, for j = 1, let D v (i,1)=1, then D h The values ​​satisfy the following relationship:

[0109]

[0110] The lateral gain difference matrix D is obtained from thish .

[0111] 109. Transform the lateral gain difference matrix and perform row-wise multiplication to obtain a lateral gain correction matrix.

[0112] In this embodiment, the goal of the lateral gain correction matrix is ​​to transform the lateral gain difference matrix so that the value of each element in the transformed matrix can be used to obtain the image after the longitudinal correction. The gain non-uniformity correction of the pixels in the corresponding column in D is the gain non-uniformity coefficient of a certain t-th column. Specifically, the image gain non-uniformity correction device of the staring type infrared detector can be used for D h The matrix undergoes a column-swapping transformation in elementary transformations, and D h The first column is exchanged with the next column in its position one by one until the original first column elements are moved to the tth column position, and the first t-1 columns of elements after the move are set to the reciprocal of the original values ​​to obtain the horizontal transformation matrix

[0113]

[0114] After that, the horizontal transformation matrix Perform row-by-row multiplication to obtain the lateral gain correction matrix The value of each element in in, Calculate using the following formula:

[0115]

[0116] but The elements of which are related to the first correction image The pixel values ​​of the corresponding rows in are multiplied by each other, and the lateral gain correction matrix is The third frame image can be The gain non-uniformity coefficients of all pixels in are corrected to the non-uniformity coefficients of the tth row

[0117] 110. Add the lateral gain correction matrix to a second FIFO queue until the lateral gain correction matrix stored in the second FIFO queue reaches a second preset number.

[0118] In this embodiment, the image gain non-uniformity correction device of the staring type infrared detector can establish a length L h (That is, the second preset number is L h ) is a first-in-first-out queue, denoted as hFIFO (i.e., the second first-in-first-out queue), which is used to store the latest L hThe image gain non-uniformity correction device of the staring infrared detector queues the determined longitudinal gain correction matrix into the hFIFO until the length of the transverse gain correction matrix in the queue reaches L h So far, the queue stores multiple lateral gain correction matrices in the order of entry.

[0119] 111. Calculate an average matrix of a second preset number of lateral gain correction matrices in a second FIFO queue to obtain an average lateral gain correction matrix.

[0120] In this embodiment, the image gain non-uniformity correction device of the staring infrared detector can calculate the average matrix of the second preset number of lateral gain correction matrices in the second first-in-first-out queue. Specifically, the average lateral gain correction matrix is ​​determined by the following formula: The values ​​of each element in:

[0121]

[0122] in, is the value of each element in the transverse and longitudinal gain correction matrix, The second preset number L h The average value of different longitudinal gain correction coefficients, It can be calculated by the following formula;

[0123]

[0124] It should be noted that whenever a new average lateral gain correction matrix is ​​calculated After that, clear all queue elements in hFIFO and wait for L h New lateral gain correction matrices are queued until the queue is full, and then the next new average lateral gain correction matrix is ​​calculated.

[0125] 112. Determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix.

[0126] In this embodiment, after determining the average longitudinal gain correction matrix and the average transverse gain correction matrix, the image gain non-uniformity correction device of the staring type infrared detector can determine the average gain correction matrix by the following formula: The value at coordinate (i, j) (that is, the average longitudinal gain correction matrix and the average lateral gain correction matrix Multiply by the index to get the average gain correction matrix The value at coordinate (i,j)):

[0127]

[0128] in, is the average gain correction matrix The value at coordinate (i,j), is the value of each element in the average longitudinal gain correction matrix, is the value of each element in the average lateral gain correction matrix, The second preset number L h The average value of different lateral gain correction coefficients, g(i,j) is the gain non-uniformity parameter;

[0129] It should be noted that in order to ensure the average vertical correction matrix value at each pixel coordinate (i, j) The numerator and mean transverse correction matrix values ​​of The denominators of just cancel each other, only when the average gain correction matrix is ​​completed once After the calculation, the new average longitudinal gain correction matrix can be calculated again using the new longitudinal gain correction matrix saved in vFIFO. Then based on the new Calculate the new average lateral gain correction matrix

[0130] It should also be noted that, since infrared non-uniformity will have a slow time drift, the bias and gain coefficients of the detector will change, so two queues are constructed in the present invention, and the average longitudinal gain correction matrix and the average transverse gain correction matrix are calculated after the number of gain correction matrices in the queue reaches a threshold, and then the average gain matrix is ​​calculated based on the average longitudinal gain correction matrix and the average transverse gain correction matrix. That is, the average gain correction matrix is ​​continuously updated with the update of the average longitudinal gain correction matrix and the average transverse gain correction matrix, and the average gain correction matrix, the average longitudinal correction matrix and the average transverse gain correction matrix correspond to each other.

[0131] 113. Perform gain non-uniformity correction on the input image according to the average gain correction matrix.

[0132] In this embodiment, the image gain non-uniformity correction device of the staring type infrared detector can perform gain non-uniformity correction on the input image through the average gain correction matrix (the average gain correction matrix is ​​the average gain correction matrix calculated at the current moment), and the input image is an infrared image after the bias non-uniformity is corrected. The bias non-uniformity correction here is similar to that in step 101. Specifically, the gain non-uniformity correction is performed on the input image through the following formula:

[0133]

[0134] Among them, w o (i,j) is the average gain correction matrix The corrected input image W o The value at coordinate (i,j), The second preset number L h The average value of different lateral gain correction coefficients, φ o (i, j) is the actual infrared radiation value of the scene corresponding to the input image.

[0135] In summary, it can be seen that in the embodiment provided by the present application, a longitudinal gain difference matrix is ​​constructed by the longitudinal difference of the infrared detector signals output by the same scene radiation in the previous and next frame images, and an average longitudinal gain correction matrix is ​​obtained by further transforming and calculating the longitudinal gain difference matrix, and the image is longitudinally corrected by the average longitudinal gain correction matrix, and a lateral gain difference matrix is ​​constructed based on the lateral difference of the infrared detector signals output by the same scene radiation in the image after the longitudinal correction, and an average lateral gain correction matrix is ​​obtained by further transforming and calculating the lateral gain difference matrix, and finally the average gain correction matrix is ​​determined according to the average longitudinal gain correction matrix and the average lateral gain correction matrix, and finally the input image is corrected for gain non-uniformity by the average gain correction matrix, so as to solve the problem of residual gain non-uniformity in the image output by the traditional infrared non-uniformity correction method based on registration, further correct the gain non-uniformity, and improve the accuracy of infrared image non-uniformity correction.

[0136] The above describes the embodiment of the present invention from the perspective of the image gain non-uniformity correction method of a staring type infrared detector. The following describes the embodiment of the present invention from the perspective of the image gain non-uniformity correction device of a staring type infrared detector.

[0137] See also Figure 2 , a virtual structural diagram of an image gain non-uniformity correction device for a staring type infrared detector according to an embodiment of the present invention, wherein the image gain non-uniformity correction device 200 for a staring type infrared detector comprises:

[0138] The determination module 201 is used to determine a first detector output signal of a first frame image after the bias non-uniformity correction, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any two adjacent infrared frames of images having longitudinal sub-pixel translation in the multiple frames after the bias non-uniformity correction;

[0139] The determination module 201 is further configured to determine the output signal of the first registration detector of the second frame image according to the output signal of the second detector and the inter-frame sub-pixel longitudinal motion translation;

[0140] A matrix construction module 202, configured to construct a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector, and the longitudinal motion translation of the sub-pixel between frames;

[0141] The transformation and multiplication module 203 is used to transform and multiply the longitudinal gain difference matrix by columns to obtain a longitudinal gain correction matrix;

[0142] An adding module 204 is used to add the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number;

[0143] A calculation module 205, configured to calculate an average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix;

[0144] A correction module 206 is used to perform longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared frames of images with lateral sub-pixel shift in the multi-frame image after the bias non-uniformity is corrected;

[0145] The determination module 201 is further configured to determine the output signal of the second registration detector of the fourth frame image according to the output signal of the fourth detector and the inter-frame sub-pixel lateral motion translation amount.

[0146] The matrix construction module 202 is further used to construct a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector and the inter-frame sub-pixel lateral motion translation;

[0147] The transformation and multiplication module 203 is further used to transform and multiply the lateral gain difference matrix row by row to obtain a lateral gain correction matrix;

[0148] The adding module 204 is further configured to add the lateral gain correction matrix to a second FIFO queue until the lateral gain correction matrix stored in the second FIFO queue reaches a second preset number;

[0149] The calculation module 205 is further configured to calculate an average matrix of the second preset number of lateral gain correction matrices in the second FIFO queue to obtain an average lateral gain correction matrix;

[0150] The determination module 201 is further configured to determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix;

[0151] The correction module 206 is further configured to perform gain non-uniformity correction on the input image according to the average gain correction matrix, wherein the input image is an infrared image after bias non-uniformity correction.

[0152] above Figure 3 The image gain non-uniformity correction device of the staring type infrared detector in the embodiment of the present invention is described from the perspective of modular functional entities. The image gain non-uniformity correction device of the staring type infrared detector in the embodiment of the present invention is described in detail from the perspective of hardware processing. Please refer to FIG. 300, which is a schematic diagram of an embodiment of the image gain non-uniformity correction device 300 of the staring type infrared detector in the embodiment of the present invention. The image gain non-uniformity correction device 300 of the staring type infrared detector includes:

[0153] An input device 301, an output device 302, a processor 303 and a memory 304 (the number of the processor 303 can be one or more, Figure 3 In some embodiments of the present invention, the input device 301, the output device 302, the processor 303 and the memory 304 may be connected via a communication bus or other means, wherein: Figure 3 The communication bus connection is taken as an example.

[0154] Wherein, by calling the operation instruction stored in the memory 304, the processor 303 is used to perform the following steps:

[0155] Determine a first detector output signal of a first frame image after correcting bias non-uniformity, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any two adjacent infrared frames of images having longitudinal sub-pixel translation in the multiple frames after correcting bias non-uniformity;

[0156] Determine the output signal of the first registration detector of the second frame image according to the output signal of the second detector and the inter-frame sub-pixel longitudinal motion translation;

[0157] Constructing a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector and the longitudinal motion translation of the sub-pixel between frames;

[0158] Transforming and column-wise multiplying the longitudinal gain difference matrix to obtain a longitudinal gain correction matrix;

[0159] Adding the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number;

[0160] Calculating an average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix;

[0161] Performing longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared images with lateral sub-pixel shift in the multi-frame image after the bias non-uniformity is corrected;

[0162] Determine the second registration detector output signal of the fourth frame image according to the fourth detector output signal and the inter-frame sub-pixel lateral motion translation

[0163] Constructing a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector and the inter-frame sub-pixel lateral motion translation;

[0164] Transforming and row-wise multiplying the lateral gain difference matrix to obtain a lateral gain correction matrix;

[0165] Adding the lateral gain correction matrix to a second first-in-first-out queue until the number of lateral gain correction matrices stored in the second first-in-first-out queue reaches a second preset number;

[0166] Calculating an average matrix of the second preset number of lateral gain correction matrices in the second FIFO queue to obtain an average lateral gain correction matrix;

[0167] Determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix;

[0168] The gain non-uniformity correction is performed on the input image according to the average gain correction matrix, and the input image is an infrared image after the bias non-uniformity correction is performed.

[0169] By calling the operation instructions stored in the memory 304, the processor 303 is also used to execute Figure 1 Any method in the corresponding embodiment.

[0170] See also Figure 4 , Figure 4 A schematic diagram of an electronic device according to an embodiment of the present invention.

[0171] like Figure 4As shown, an embodiment of the present invention provides an electronic device, including a memory 410, a processor 420, and a computer program 411 stored in the memory 410 and executable on the processor 420. When the processor 420 executes the computer program 411, the following steps are implemented:

[0172] Determine a first detector output signal of a first frame image after correcting bias non-uniformity, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any two adjacent infrared frames of images having longitudinal sub-pixel translation in the multiple frames after correcting bias non-uniformity;

[0173] Determine the output signal of the first registration detector of the second frame image according to the output signal of the second detector and the inter-frame sub-pixel longitudinal motion translation;

[0174] Constructing a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector and the longitudinal motion translation of the sub-pixel between frames;

[0175] Transforming and column-wise multiplying the longitudinal gain difference matrix to obtain a longitudinal gain correction matrix;

[0176] Adding the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number;

[0177] Calculating an average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix;

[0178] Performing longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared images with lateral sub-pixel shift in the multi-frame image after the bias non-uniformity is corrected;

[0179] Determine the second registration detector output signal of the fourth frame image according to the fourth detector output signal and the inter-frame sub-pixel lateral motion translation

[0180] Constructing a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector and the inter-frame sub-pixel lateral motion translation;

[0181] Transforming and row-wise multiplying the lateral gain difference matrix to obtain a lateral gain correction matrix;

[0182] Adding the lateral gain correction matrix to a second first-in-first-out queue until the number of lateral gain correction matrices stored in the second first-in-first-out queue reaches a second preset number;

[0183] Calculating an average matrix of the second preset number of lateral gain correction matrices in the second FIFO queue to obtain an average lateral gain correction matrix;

[0184] Determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix;

[0185] The gain non-uniformity correction is performed on the input image according to the average gain correction matrix, and the input image is an infrared image after the bias non-uniformity correction is performed.

[0186] In the specific implementation process, when the processor 420 executes the computer program 411, it can achieve Figure 1 Any implementation manner in the corresponding embodiments.

[0187] Since the electronic device introduced in this embodiment is a device used to implement an image gain non-uniformity correction device of a staring type infrared detector in the embodiment of the present invention, based on the method introduced in the embodiment of the present invention, technical personnel in this field can understand the specific implementation mode of the electronic device of this embodiment and its various variations. Therefore, how the electronic device implements the method in the embodiment of the present invention will not be introduced in detail here. As long as the equipment used by technical personnel in this field to implement the method in the embodiment of the present invention belongs to the scope of protection of the present invention.

[0188] Please refer to Figure 500, which is a schematic diagram of an embodiment of a computer-readable storage medium provided by an embodiment of the present invention.

[0189] As shown in Figure 500, an embodiment of the present invention further provides a computer-readable storage medium 500, on which a computer program 511 is stored. When the computer program 511 is executed by a processor, the following steps are implemented:

[0190] Determine a first detector output signal of a first frame image after correcting bias non-uniformity, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any two adjacent infrared frames of images having longitudinal sub-pixel translation in the multiple frames after correcting bias non-uniformity;

[0191] Determine the output signal of the first registration detector of the second frame image according to the output signal of the second detector and the inter-frame sub-pixel longitudinal motion translation;

[0192] Constructing a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector and the longitudinal motion translation of the sub-pixel between frames;

[0193] Transforming and column-wise multiplying the longitudinal gain difference matrix to obtain a longitudinal gain correction matrix;

[0194] Adding the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number;

[0195] Calculating an average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix;

[0196] Performing longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared images with lateral sub-pixel shift in the multi-frame image after the bias non-uniformity is corrected;

[0197] Determine the second registration detector output signal of the fourth frame image according to the fourth detector output signal and the inter-frame sub-pixel lateral motion translation

[0198] Constructing a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector and the inter-frame sub-pixel lateral motion translation;

[0199] Transforming and row-wise multiplying the lateral gain difference matrix to obtain a lateral gain correction matrix;

[0200] Adding the lateral gain correction matrix to a second first-in-first-out queue until the number of lateral gain correction matrices stored in the second first-in-first-out queue reaches a second preset number;

[0201] Calculating an average matrix of the second preset number of lateral gain correction matrices in the second FIFO queue to obtain an average lateral gain correction matrix;

[0202] Determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix;

[0203] The gain non-uniformity correction is performed on the input image according to the average gain correction matrix, and the input image is an infrared image after the bias non-uniformity correction is performed.

[0204] In the specific implementation process, the computer program 511 is executed by the processor to achieve Figure 1 Any implementation manner in the corresponding embodiments.

[0205] The embodiment of the present invention also provides a computer program product, which includes computer software instructions. When the computer software instructions are executed on a processing device, the processing device executes the following Figure 1 The process in the corresponding embodiment.

[0206] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0207] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for correcting image gain non-uniformity of a staring infrared detector, characterized in that: include: Determine a first detector output signal of a first frame image after correcting bias non-uniformity, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any two adjacent infrared frames of images having longitudinal sub-pixel translation in the multiple frames after correcting bias non-uniformity; Determine the output signal of the first registration detector of the second frame image according to the output signal of the second detector and the inter-frame sub-pixel longitudinal motion translation; Constructing a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector and the longitudinal motion translation of the sub-pixel between frames; Transforming and column-wise multiplying the longitudinal gain difference matrix to obtain a longitudinal gain correction matrix; Adding the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number; Calculating an average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix; Performing longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared images with lateral sub-pixel shift in the multi-frame image after the bias non-uniformity is corrected; Determine the second registration detector output signal of the fourth frame image according to the fourth detector output signal and the inter-frame sub-pixel lateral motion translation Constructing a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector and the inter-frame sub-pixel lateral motion translation; Transforming and row-wise multiplying the lateral gain difference matrix to obtain a lateral gain correction matrix; Adding the lateral gain correction matrix to a second first-in-first-out queue until the number of lateral gain correction matrices stored in the second first-in-first-out queue reaches a second preset number; Calculating an average matrix of the second preset number of lateral gain correction matrices in the second FIFO queue to obtain an average lateral gain correction matrix; Determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix; The gain non-uniformity correction is performed on the input image according to the average gain correction matrix, and the input image is an infrared image after the bias non-uniformity correction is performed.

2. The method according to claim 1, characterized in that: The determining of the first detector output signal of the first frame image after the bias non-uniformity correction comprises: The detector output signal is determined by the following formula: ； in, is the output signal of the first detector, , , is the resolution of the first frame image, is the actual infrared radiation value of the scene corresponding to the first frame image, is the gain non-uniformity parameter.

3. The method according to claim 1, characterized in that The step of constructing a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector and the longitudinal motion translation of the sub-pixel between frames comprises: The values ​​of each element in the longitudinal gain difference matrix are determined by the following formula: ； in, is the longitudinal gain difference matrix In coordinates The value of is the output signal of the first detector, output signal for the first registration detector, is the sub-pixel longitudinal motion translation between frames; ； or, ； in, is the gain coefficient after ideal longitudinal translation; The longitudinal gain difference matrix is ​​constructed based on the values ​​of each element in the longitudinal gain difference matrix by the following formula: 。 4. The method according to claim 1, characterized in that The step of transforming the gain difference matrix and multiplying the matrix by columns to obtain a longitudinal gain correction matrix comprises: The gain difference matrix is ​​transformed by the following formula to obtain the longitudinal transformation matrix: in, is the longitudinal transformation matrix; The longitudinal gain correction matrix is ​​obtained by multiplying the longitudinal transformation matrix column by column using the following formula: The values ​​of each element: ； in, is the longitudinal gain correction matrix The value of each element in , is the longitudinal gain correction matrix The non-uniformity coefficient of the sth row in .

5. The method according to claim 1, characterized in that Calculating the average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain the average longitudinal gain correction matrix includes: The values ​​of each element in the average longitudinal gain correction matrix are determined by the following formula: ； in, is the value of each element in the average longitudinal gain correction matrix, The first preset number The average value of different longitudinal gain correction coefficients, ; The average longitudinal gain correction matrix is ​​constructed according to the values ​​of each element.

6. The method according to claim 1, characterized in that The step of performing longitudinal correction on the third frame image and the fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image comprises: The output signal of the third detector is determined by the following formula: ； in, is the third detector output signal, The first preset number The average value of different longitudinal gain correction coefficients, is the real infrared radiation value of the scene corresponding to the third frame image; The output signal of the fourth detector is determined by the following formula: ； in, is the fourth detector output signal, The first preset number The average value of different longitudinal gain correction coefficients, is the actual infrared radiation value of the scene corresponding to the fourth frame image.

7. The method according to claim 1, characterized in that Determining the average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix comprises: The average gain correction matrix is ​​determined by the following formula In coordinates Values: ； in, is the average gain correction matrix In coordinates The value of is the value of each element in the average longitudinal gain correction matrix, is the value of each element in the average lateral gain correction matrix, The second preset number The average value of different lateral gain correction coefficients, is the gain non-uniformity parameter; According to the average gain correction matrix In coordinates The value of determines the average gain correction matrix.

8. The method according to claim 1, characterized in that The performing gain non-uniformity correction on the input image according to the average gain correction matrix comprises: The gain non-uniformity correction of the input image is performed by the following formula: ； in, is the average gain correction matrix The input image after correction In coordinates The value of The second preset number The average value of different lateral gain correction factors.

9. An image gain non-uniformity correction device for a staring infrared detector, characterized in that: include: A determination module, used to determine a first detector output signal of a first frame image after bias non-uniformity correction, a second detector output signal of a second frame image, and an amount of sub-pixel longitudinal motion translation between frames, wherein the first frame image and the second frame image are any two adjacent infrared frames of images having longitudinal sub-pixel translation in a plurality of frames after bias non-uniformity correction; The determination module is further used to determine the output signal of the first registration detector of the second frame image according to the output signal of the second detector and the inter-frame sub-pixel longitudinal motion translation; A matrix construction module, used for constructing a longitudinal gain difference matrix according to the output signal of the first detector, the output signal of the first registration detector and the longitudinal motion translation of the sub-pixel between frames; A transformation and multiplication module, used for transforming and multiplying the longitudinal gain difference matrix by columns to obtain a longitudinal gain correction matrix; An adding module, configured to add the longitudinal gain correction matrix to a first FIFO queue until the number of longitudinal gain correction matrices stored in the first FIFO queue reaches a first preset number; A calculation module, used for calculating an average matrix of the first preset number of longitudinal gain correction matrices in the first FIFO queue to obtain an average longitudinal gain correction matrix; a correction module, used for performing longitudinal correction on a third frame image and a fourth frame image according to the average longitudinal gain correction matrix to obtain a third detector output signal of the third frame image and a fourth detector output signal of the fourth frame image, wherein the third frame image and the fourth frame image are any two adjacent infrared images with lateral sub-pixel shift in the multi-frame image after the bias non-uniformity is corrected; The determination module is further used to determine the output signal of the second registration detector of the fourth frame image according to the output signal of the fourth detector and the inter-frame sub-pixel lateral motion translation amount. The matrix construction module is further used to construct a lateral gain difference matrix according to the output signal of the third detector, the output signal of the second registration detector and the inter-frame sub-pixel lateral motion translation; The transformation and multiplication module is further used to transform and multiply the lateral gain difference matrix row by row to obtain a lateral gain correction matrix; The adding module is further used to add the lateral gain correction matrix to a second first-in-first-out queue until the lateral gain correction matrix stored in the second first-in-first-out queue reaches a second preset number; The calculation module is further used to calculate the average matrix of the second preset number of lateral gain correction matrices in the second FIFO queue to obtain an average lateral gain correction matrix; The determination module is further configured to determine an average gain correction matrix according to the average longitudinal gain correction matrix and the average lateral gain correction matrix; The correction module is further used to perform gain non-uniformity correction on the input image according to the average gain correction matrix, and the input image is an infrared image after the bias non-uniformity is corrected.

10. A computer-readable storage medium having a computer management program stored thereon, characterized in that: When the computer management program is executed by a processor, the gain non-uniformity correction method for infrared images as claimed in any one of claims 1 to 8 is implemented.

Images