Infrared non-uniformity noise correction method and system

Abstract

Disclosed in the present invention are an infrared non-uniformity noise correction method and system. The method comprises the following steps: determining whether an infrared imaging device has moved, if yes, acquiring non-uniformity correction values and non-uniformity correction value change amounts of pixel points on the basis of a gradient matrix and a truth value matrix, and performing corresponding updating processing on the pixel points; completing iterative calculation of a bias matrix of the current image frame on the basis of the updated pixel points; and performing non-uniformity correction on the current image frame on the basis of the bias matrix. The present invention first extracts a clear non-uniform noise pattern by means of the bias matrix, and then corrects the non-uniform noise by means of the bias matrix, so that the non-uniform noise in the infrared image can be significantly reduced, and the details of a target object are not lost, and thus, the image contrast and quality are greatly improved.

Description

Infrared non-uniform noise correction method and system Technical Field

[0001] The present invention relates to the field of infrared imaging technology, and in particular to an infrared non-uniform noise correction method and system. Background Art

[0002] When infrared thermal imaging equipment is in use, environmental noise, focal temperature, and other variations in the equipment itself can cause noise, further impacting image quality due to non-uniform noise. Unlike dynamically changing Gaussian noise, non-uniform noise is a fixed feature that appears on the image, significantly impacting the viewing experience.

[0003] In order to eliminate this type of non-uniform noise, most infrared thermal imaging devices are equipped with a shutter for shutter compensation. However, if the shutter fails, it will not work properly, and an excessively high shutter frequency will reduce the service life of the device.

[0004] Therefore, technical solutions have been developed to correct the above-mentioned non-uniform noise on the image algorithm side. For example, by estimating the true value of multiple recent frames of images, establishing a noise model for multiple original images and true value images, and fitting the noise model parameters to fit the equation for the non-uniform noise, the non-uniform noise can be removed. However, this solution has a very large computational load and cannot be successfully deployed on a suitable platform. It can only be used in offline simulations, and its actual application scope is limited. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides an infrared non-uniform noise correction method and system, which extracts a clear non-uniform noise pattern through a bias matrix and then corrects the non-uniform noise through the bias matrix. This can significantly reduce the non-uniform noise in the infrared image without losing the details of the target object, thereby greatly improving the image contrast and quality.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] In one aspect, a method for correcting infrared non-uniform noise is provided, comprising the following steps:

[0008] Determine whether the infrared imaging device has moved. If so, obtain the non-uniformity correction value of the pixel point and the change in the non-uniformity correction value based on the gradient matrix and the true value matrix, and perform corresponding update processing on the pixel point;

[0009] Complete the iterative calculation of the bias matrix of the current frame image according to the updated pixel points;

[0010] Perform non-uniformity correction on the current frame image according to the bias matrix.

[0011] Preferably, the gradient matrix is ​​obtained by formula (1):

[0012] Among them, GX i,j is the gradient matrix; x is the grayscale value of the pixel in the current frame image; i and j are the horizontal and vertical coordinates of the pixel in the current frame image respectively; k is the distance between the current pixel and the center pixel.

[0013] Preferably, the truth matrix is ​​obtained by formula (2):

[0014] Among them, T i,j is a true value matrix; x is the grayscale value of the pixel in the current frame image; i and j are the horizontal and vertical coordinates of the pixel in the current frame image respectively; M and N are the number of columns and rows of pixels in the local image centered on the pixel (i, j) respectively; p is the distance between the current pixel and the center pixel.

[0015] Preferably, obtaining the non-uniformity correction value of the pixel point and the variation of the non-uniformity correction value according to the gradient matrix and the true value matrix includes the following steps:

[0016] The non-uniformity correction value change c of the pixel point is obtained by formula (3):

[0017] By o k =o k-1 -c gets the non-uniformity correction value of the pixel point, o k-1 、o k are the non-uniformity correction values ​​of the current pixel in the previous frame image and the current frame image respectively.

[0018] Preferably, performing corresponding update processing on the pixel points includes the following steps:

[0019] When o k When the non-uniformity correction value threshold o0 and c< the non-uniformity correction value variation threshold c0, the non-uniformity correction value o of the current pixel in the previous frame image is k-1 Replace with o k , to complete the update of the current pixel.

[0020] Preferably, the value range of the non-uniformity correction value threshold o0 is [-30, 30], and the value range of the non-uniformity correction value variation threshold c0 is [-10, 10].

[0021] Preferably, completing the iterative calculation of the bias matrix O of the current frame image according to the updated pixel points includes the following steps:

[0022] The iterative calculation of the bias matrix O is completed through formula (4):

[0023] in, are the bias matrices of the previous frame image and the current frame image respectively; λ is the iteration step size.

[0024] Preferably, performing non-uniformity correction on the current frame image according to the bias matrix O includes the following steps:

[0025] The non-uniformity correction of the current frame image is performed using formula (5):

[0026] in, is the corrected current frame image; when the infrared imaging device does not move, is the bias matrix O of the current frame image without iterative calculation; when the infrared imaging device moves, It is the bias matrix O of the current frame image after iterative calculation in step S5.

[0027] On the other hand, there is also provided an infrared non-uniformity noise correction system, comprising:

[0028] A gradient matrix calculation module is used to obtain a current gradient matrix according to a current frame image;

[0029] A truth matrix calculation module is used to obtain a current truth matrix based on the current frame image;

[0030] A movement determination module, which is used to determine whether the infrared imaging device moves;

[0031] A pixel update module is used to obtain the non-uniformity correction value of the pixel point and the change in the non-uniformity correction value of the pixel point during the inter-frame iteration process when the infrared imaging device moves, and complete the update processing of the current pixel point based on the non-uniformity correction value of the current pixel point and the change in the non-uniformity correction value;

[0032] A bias matrix iteration module, which is used to complete the iterative calculation of the bias matrix of the current frame image according to the updated pixel points;

[0033] The non-uniformity correction module is used to perform non-uniformity correction on the current frame image according to the bias matrix.

[0034] Preferably, the gradient matrix calculation module, the truth matrix calculation module, the movement judgment module, the pixel update module, the bias matrix iteration module, the non-uniformity correction module and the storage module are integrated on a chip.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention gradually extracts clear non-uniform noise patterns through a bias matrix, and then corrects the non-uniform noise through the bias matrix, which can significantly reduce the non-uniform noise in infrared images without losing the details of the target object, thereby greatly improving the image contrast and quality. In addition, the method can be deployed on FPGA and ARM platforms, with controllable computational complexity and no need to use too many frame buffers or complex filtering operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] FIG1 is a flowchart of the steps of the mid-infrared non-uniformity noise correction method of the present invention;

[0038] Figure 2 shows an infrared image containing non-uniform noise;

[0039] FIG3 is an infrared image before and after processing by the present invention;

[0040] FIG4 is a schematic structural diagram of the infrared non-uniformity noise correction system of the present invention. DETAILED DESCRIPTION

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0042] Example 1:

[0043] This embodiment provides an infrared non-uniform noise correction method, which includes the following steps:

[0044] S1. Parameter initialization to obtain an initial bias matrix O, an initial gradient matrix GX, an initial true value matrix T, and an initial historical correction frame Lc, wherein the initial bias matrix O, the initial gradient matrix GX, the initial true value matrix T, and the initial historical correction frame Lc are all 0;

[0045] S2. Obtaining a gradient matrix GX and a true value matrix T based on a current frame image that has undergone background correction (which may be an infrared image such as Y16 data); wherein the background correction may be implemented by a two-point correction algorithm, for example;

[0046] Specifically, in this embodiment, the gradient matrix GX is obtained by formula (1):

[0047] Among them, GX i,jThat is, the gradient matrix GX; x is the grayscale value of the pixel in the current frame image; i and j are the horizontal and vertical coordinates of the pixel in the current frame image respectively; k is the distance between the current pixel and the center pixel, which is used to calculate the gradient change value of the adjacent pixels. For example, in this embodiment, k=1;

[0048] The truth matrix T is obtained by formula (2):

[0049] Among them, T i,j is the true value matrix T; x is the grayscale value of the pixel point of the current frame image; i and j are the horizontal and vertical coordinates of the pixel point of the current frame image respectively; M and N are the number of columns and rows of the pixel points of the local image centered on the pixel point (i, j), respectively. The values ​​can be set according to actual needs, and M and N can be equal or different; for example, when M=5 and N=5, it means that the number of columns and rows of the pixel points of the local image centered on the pixel point (i, j) are both 5. Furthermore, in order to save resources, M=N=3 in this embodiment; p is the distance value between the current pixel point and the center pixel point, which is used to calculate the filter value. The filter value is used as the true value for subsequent calculations, and p is a positive integer. For example, in this embodiment, p=2;

[0050] Thus, the gradient matrix GX can be obtained by the degree of change of the four-directional gradient to fix the non-uniform noise and extract the gradient information of the noise. At the same time, the result of the mean filter is used as the true value matrix T to remove sharp noise and achieve image smoothing.

[0051] S3, determining whether the infrared imaging device has moved, if so, executing step S4, if not, executing step S6;

[0052] Specifically, whether the infrared imaging device has moved can be determined by a sensor such as an IMU (Inertial Measurement Unit), for example, by judging by angular velocity, or by determining whether the infrared imaging device has moved by scene content contained in adjacent frame images. Both of these techniques are prior art and will not be further described.

[0053] S4, obtaining the non-uniformity correction value variation c of the current pixel point in the current frame image during the inter-frame iteration process and the non-uniformity correction value o of the current pixel point k , and according to the non-uniformity correction value change c of the current pixel point and the non-uniformity correction value o k Complete the update process of the current pixel;

[0054] Among them, k =o k-1 -c,o k-1 、o kThe non-uniformity correction value of the current pixel in the previous frame image (i.e., the k-1 frame image, also known as the historical correction frame Lc, which is also an infrared image such as Y16 data) and the current frame image (i.e., the k-th frame image);

[0055] The non-uniformity correction value change c of the current pixel point is obtained by formula (3):

[0056] Furthermore, completing the update process of the current pixel includes the following steps:

[0057] When o k When the non-uniformity correction value threshold o0 and c< the non-uniformity correction value variation threshold c0, the non-uniformity correction value o of the current pixel in the previous frame image is k-1 Replace with o k , to complete the update of the current pixel point; the non-uniformity correction value threshold o0 and the non-uniformity correction value change threshold c0 can be set according to actual needs. For example, in this embodiment, the value range of the non-uniformity correction value threshold o0 is [-30, 30], and the value range of the non-uniformity correction value change threshold c0 is [-10, 10];

[0058] When o≥non-uniformity correction value threshold, and / or c≥non-uniformity correction value change threshold, it means that the pixel value of the pixel point has changed significantly between adjacent frames, and this change is mainly caused by the change of scene gray value and has nothing to do with image non-uniformity noise. Therefore, there is no need to update the current pixel point, that is, the non-uniformity correction value o of the current pixel point in the previous frame image is k-1 No need to replace with o k When performing non-uniformity correction on subsequent images, the non-uniformity correction value of the current pixel in the previous frame of image is still used. k-1 , to reduce the impact of artifacts;

[0059] Repeat the above steps until every pixel in the current frame image is updated;

[0060] S5. Complete the iterative calculation of the bias matrix O of the current frame image according to the updated pixel points;

[0061] Specifically, in this embodiment, the iterative calculation of the bias matrix O is completed by formula (4):

[0062] in, are the bias matrices of the previous frame image (i.e., the k-1th frame image) and the current frame image (i.e., the kth frame image), respectively; λ is the iteration step size, and its value range is [0,1];

[0063] And S6, performing non-uniformity correction on the current frame image according to the bias matrix O;

[0064] Specifically, this embodiment performs non-uniformity correction on the current frame image using formula (5):

[0065] in, This is the corrected current frame image; when the infrared imaging device does not move, is the bias matrix O of the current frame image without iterative calculation; when the infrared imaging device moves, The bias matrix O of the current frame image after iterative calculation in step S5;

[0066] At the same time, the historical correction frame Lc is updated to the corrected current frame image to serve as the previous frame image when performing non-uniformity correction on the next burst of images. After completing the non-uniformity correction of several frames (e.g., 10,000 frames), the bias matrix O used in the most recent correction is saved.

[0067] Therefore, this embodiment extracts noise features through four-directional gradient changes and mean filtering, and further uses the total variation iterative solution optimization method to gradually extract a clear non-uniform noise form through the bias matrix O. As can be seen from Figure 2, non-uniform noise mostly appears as spots and horizontal and vertical stripes. On this basis, the non-uniform noise is corrected by the bias matrix O. As can be seen from Figure 3, compared with the image in the left column that has not been corrected for non-uniform noise, after processing by the correction method of this embodiment, the non-uniform noise in the infrared image (i.e., the image in the right column of Figure 3) is significantly reduced without losing the details of the target object, thereby greatly improving the image contrast and quality.

[0068] Example 2:

[0069] This embodiment provides an infrared non-uniform noise correction system, which can implement the infrared non-uniform noise correction method described in Example 1. The system includes:

[0070] Parameter initialization module 1, which is used to perform parameter initialization settings to obtain the initial bias matrix O, the initial gradient matrix GX, the initial true value matrix T and the initial historical correction frame Lc;

[0071] Gradient matrix calculation module 2, which is used to obtain the current gradient matrix GX according to the current frame image after background correction, and its process is the same as step S2;

[0072] A truth matrix calculation module 3 is used to obtain a current truth matrix T based on the background-corrected current frame image, and the process is the same as step S2;

[0073] A movement determination module 4 is used to determine whether the infrared imaging device moves;

[0074] Pixel update module 5 is used to obtain the non-uniformity correction value o of the current pixel point in the current frame image and the non-uniformity correction value change c of the current pixel point during the inter-frame iteration process when the infrared imaging device moves, and complete the update processing of the current pixel point based on the non-uniformity correction value o of the current pixel point and the non-uniformity correction value change c of the current pixel point. The process is the same as step S4;

[0075] A bias matrix iteration module 6 is used to perform iterative calculation of the bias matrix O of the current frame image according to the updated pixel points;

[0076] A non-uniformity correction module 7 is used to perform non-uniformity correction on the current frame image according to the bias matrix O, and the process is the same as step S6;

[0077] And a saving module 8 is used to save the bias matrix O used in the most recent correction after completing the non-uniformity correction of several frames of images.

[0078] Furthermore, the parameter initialization module 1, the gradient matrix calculation module 2, the true value matrix calculation module 3, the motion judgment module 4, the pixel update module 5, the bias matrix iteration module 6, the non-uniformity correction module 7 and the storage module 8 are integrated on a chip. For example, in this embodiment, the chip includes an FPGA chip or an ARM chip.

[0079] To sum up, the present application extracts noise features through four-directional gradient changes and mean filtering, and further uses the total variation iterative solution optimization method to gradually extract clear non-uniform noise forms through the bias matrix, and then corrects the non-uniform noise through the bias matrix, which can significantly reduce the non-uniform noise in the infrared image without losing the details of the target object, thereby greatly improving the image contrast and quality. Furthermore, the computational complexity of the method in the present application is controllable, and there is no need to use too many frame buffers or complex filtering operations, and it can be implemented on both online FPGA and ARM platforms.

[0080] It should be noted that the technical features in the above-mentioned embodiments 1 to 2 can be combined in any way, and the technical solutions formed by the combination all belong to the scope of protection of this application. In this article, terms such as "including", "comprising" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or equipment. In the absence of further restrictions, the elements defined by the sentence "including a..." do not exclude the presence of other identical elements in the process, method, article or equipment including the elements.

[0081] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for correcting infrared non-uniform noise, characterized in that: The steps include: Determine whether the infrared imaging device has moved. If so, obtain the non-uniformity correction value of the pixel point and the change in the non-uniformity correction value based on the gradient matrix and the true value matrix, and perform corresponding update processing on the pixel point; Complete the iterative calculation of the bias matrix of the current frame image according to the updated pixel points; Perform non-uniformity correction on the current frame image according to the bias matrix.

2. The infrared non-uniform noise correction method according to claim 1, wherein: The gradient matrix is ​​obtained by formula (1): Among them, GX i,j is the gradient matrix; x is the grayscale value of the pixel in the current frame image; i and j are the horizontal and vertical coordinates of the pixel in the current frame image respectively; k is the distance between the current pixel and the center pixel.

3. The infrared non-uniform noise correction method according to claim 1, wherein: The truth matrix is ​​obtained by formula (2): Among them, T i,j is a true value matrix; x is the grayscale value of the pixel in the current frame image; i and j are the horizontal and vertical coordinates of the pixel in the current frame image respectively; M and N are the number of columns and rows of pixels in the local image centered on the pixel (i, j) respectively; p is the distance between the current pixel and the center pixel.

4. The infrared non-uniformity noise correction method according to any one of claims 1 to 3, characterized in that: Obtaining the non-uniformity correction value of the pixel point and the variation of the non-uniformity correction value according to the gradient matrix and the true value matrix includes the following steps: The non-uniformity correction value change c of the pixel point is obtained by formula (3): By o k =o k-1 -c gets the non-uniformity correction value of the pixel point, o k-1 、o k are the non-uniformity correction values ​​of the current pixel in the previous frame image and the current frame image respectively.

5. The infrared non-uniformity noise correction method according to claim 4, wherein: The corresponding update processing of the pixel points includes the following steps: When o k When the non-uniformity correction value threshold o0 and c< the non-uniformity correction value variation threshold c0, the non-uniformity correction value o of the current pixel in the previous frame image is k-1 Replace with o k , to complete the update of the current pixel.

6. The infrared non-uniform noise correction method according to claim 5, wherein: The value range of the non-uniformity correction value threshold o0 is [-30, 30], and the value range of the non-uniformity correction value variation threshold c0 is [-10, 10].

7. The infrared non-uniform noise correction method according to claim 4, wherein: The iterative calculation of the bias matrix O of the current frame image is completed according to the updated pixel points, including the following steps: The iterative calculation of the bias matrix O is completed through formula (4): in, are the bias matrices of the previous frame image and the current frame image respectively; λ is the iteration step size.

8. The infrared non-uniformity noise correction method according to claim 7, wherein: Performing non-uniformity correction on the current frame image according to the bias matrix O includes the following steps: The non-uniformity correction of the current frame image is performed using formula (5): in, is the corrected current frame image; when the infrared imaging device does not move, is the bias matrix O of the current frame image without iterative calculation; when the infrared imaging device moves, It is the bias matrix O of the current frame image after iterative calculation in step S5.

9. An infrared non-uniformity noise correction system, characterized in that: include: A gradient matrix calculation module is used to obtain a current gradient matrix according to a current frame image; A truth matrix calculation module is used to obtain a current truth matrix based on the current frame image; A movement determination module, which is used to determine whether the infrared imaging device moves; A pixel update module is used to obtain the non-uniformity correction value of the pixel point and the change in the non-uniformity correction value of the pixel point during the inter-frame iteration process when the infrared imaging device moves, and complete the update processing of the current pixel point based on the non-uniformity correction value of the current pixel point and the change in the non-uniformity correction value; A bias matrix iteration module, which is used to complete the iterative calculation of the bias matrix of the current frame image according to the updated pixel points; The non-uniformity correction module is used to perform non-uniformity correction on the current frame image according to the bias matrix.

10. The infrared non-uniformity noise correction system according to claim 9, wherein: The gradient matrix calculation module, the true value matrix calculation module, the movement judgment module, the pixel update module, the bias matrix iteration module, the non-uniformity correction module and the storage module are integrated on a chip.

Images