Image processing method, device and medium based on shutterless non-uniformity correction

Abstract

The application provides an image processing method, device and medium based on shutterless non-uniformity correction, wherein when a current target scene is in a static state, only a current frame original image output by an infrared detector to a current frame background image is subjected to non-uniformity calibration correction; and when the current target scene is in a motion state, the current frame original image is subjected to non-uniformity calibration correction first to obtain a calibrated original image, and then the calibrated original image is subjected to scene-based correction by using a current scene correction matrix. Therefore, the image processing method can obtain a high-quality target infrared image without a shutter, can reduce the product size and weight of an infrared device in which the infrared detector is located, and can effectively save manufacturing cost.

Description

[0001] Technology Neighborhood

[0002] This application relates to the field of infrared image processing technology, and in particular to an image processing method, device and medium based on shutterless non-uniformity correction. Background Technology

[0003] With the advancement of infrared technology, infrared devices have been widely used in military and civilian fields. However, current uncooled infrared devices generally have a shutter, mainly for two reasons: First, due to the inherent properties of the infrared detector, infrared images are non-uniform, requiring real-time shutter correction to improve image quality; second, if the infrared energy received by the detector is too strong (e.g., when the sun is too strong), the detector's pixels can be damaged, necessitating shutter closure to protect the detector when excessive infrared energy is detected.

[0004] Therefore, uncooled infrared devices are generally equipped with shutter mechanisms to correct the non-uniformity of infrared images and protect the infrared detector. Adding a shutter mechanism not only increases the size and cost of the infrared device, but also causes the infrared image to freeze at the moment the shutter is opened, resulting in a choppy viewing experience. Furthermore, some applications, such as automotive applications, require smooth, uninterrupted images, which limits the further application of infrared devices. Summary of the Invention

[0005] To at least partially improve the technical problems existing in the prior art, this application provides a low-cost and high-quality image processing method, apparatus and computer-readable storage medium based on shutterless non-uniformity correction.

[0006] According to a first aspect of the embodiments of this application, an image processing method based on shutterless non-uniformity correction is provided, comprising:

[0007] The original image of the current frame output by the infrared detector is corrected based on the background image of the current frame to obtain the calibration correction image of the current frame;

[0008] Determine whether the target scene detected by the infrared detector is currently in motion;

[0009] If the target scene is currently in motion, the current frame calibration correction image is corrected based on the current frame scene correction matrix to obtain the current frame scene correction image, and the current frame scene correction image is processed to obtain the target infrared image.

[0010] If the target scene is currently in a static state, the current frame calibration correction image is processed to obtain a target infrared image. According to a second aspect of the embodiments of the present application, an infrared device is provided, comprising a memory and a processor, wherein the memory stores a computer program executable by the processor, and the computer program is executed by the processor to implement the image processing method.

[0011] According to a third aspect of the embodiments of the present application, a computer readable storage medium is provided, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the image processing method.

[0012] As can be seen from the above, the image processing method based on shutterless non-uniformity correction provided by some embodiments of the present application, in which, when the current target scene is in a static state, only the current frame background image is used to perform non-uniformity calibration correction on the current frame original image, and the current frame calibration correction image obtained after the non-uniformity calibration correction is taken as the image to be processed by the rear-end image processing algorithm, and when the current target scene is in a motion state, the current frame original image is first subjected to non-uniformity calibration correction using the current frame background image to eliminate most of the non-uniformity caused by temperature drift. Then, the current frame original image subjected to the calibration correction is subjected to scene-based correction based on the current frame scene correction matrix to obtain a current frame scene correction image. Therefore, the image processing method does not need to use a shutter to perform non-uniformity correction on the original image output by the infrared detector, which can reduce the product size and weight of the infrared device in which the infrared detector is located, and effectively saves the manufacturing cost. In addition, the image processing method combines the non-uniformity calibration method based on calibration and the non-uniformity correction method based on scene to perform non-uniformity correction on the original image output by the infrared detector when the target scene is currently in a motion state, which effectively improves the image quality of the target infrared image. BRIEF DESCRIPTION OF DRAWINGS

[0013] The accompanying drawings are included to provide a further understanding of the application, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the present application and, together with the description, serve to explain the principles of the present application. In the drawings:

[0014] Figure 1 A flowchart of the image processing method based on shutterless non-uniformity correction provided by some embodiments of the present application is shown in the figure.

[0015] Figure 2 An original infrared image without non-uniformity correction is shown in the figure.

[0016] Figure 3The target infrared image obtained after the non-uniformity correction processing of the image processing method provided based on some embodiments of the present application;

[0017] Figure 4 The flowchart for judging the motion state of the target scene in the processing method provided by the embodiments of the present application;

[0018] Figure 5 The flowchart for obtaining the non-uniformity noise matrix in the image processing method provided by the embodiments of the present application;

[0019] Figure 6 The flowchart for obtaining the scene correction matrix in the image processing method provided by the embodiments of the present application;

[0020] Figure 7 The flowchart for performing the burn correction in the image processing method provided by the embodiments of the present application;

[0021] Figure 8 The flowchart for determining the burn correction matrix of the current frame in the image processing method provided by the embodiments of the present application;

[0022] Figure 9 The flowchart for obtaining the scene correction image of the current frame in the image processing method provided by the embodiments of the present application;

[0023] Figure 10 The flowchart for judging whether the current frame original image has the burn point of the current frame in the image processing method provided by the embodiments of the present application;

[0024] Figure 11 The calibration correction flowchart in the image processing method provided based on some embodiments of the present application;

[0025] Figure 12 The flowchart of the image processing method provided based on some embodiments of the present application;

[0026] Figure 13 The structure diagram of the infrared image processing device based on the non-shutter non-uniformity correction provided based on some embodiments of the present application;

[0027] Figure 14 The structure diagram of the infrared device provided based on some embodiments of the present application;

[0028] Figure 15 The functional module diagram of the infrared device provided based on some embodiments of the present application. DETAILED DESCRIPTION

[0029] The technical solutions of the present application are further described in detail below in combination with the drawings and specific embodiments of the present application.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of implementations of the application. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items. In the description herein, the meaning of "a," "an," and "the" includes plural referents unless otherwise clear from the content.

[0031] In the following description, references are made to "some embodiments" that describe a subset of possible embodiments, but it is to be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other as long as there is no conflict.

[0032] At present, there are mainly two types of non-uniformity correction methods for shutter-independent infrared images: the first type is based on calibration, and the second type is based on scene non-uniformity correction. The present inventors found in the research and development process that the method based on calibration is more to eliminate the influence of temperature drift, but cannot solve the non-uniformity caused by scene targets and the non-uniformity caused by aging of infrared detectors, while the scene-based non-uniformity correction needs to be corrected by motion. Therefore, some embodiments of the present application provide an image processing method based on shutterless non-uniformity correction, so as to realize non-uniformity correction of infrared images without using shutter in both moving and static scenes. For details, please refer to Figure 1 Fig. 1 shows a flowchart of an image processing method based on shutterless non-uniformity correction according to some embodiments of the present application. In some embodiments, the image processing method provided by the present application includes S02, S04, S06 and S08, and the specific description of each step is as follows.

[0033] S02: correcting the current frame original image output by the infrared detector based on the current frame background image to obtain a current frame calibration correction image.

[0034] The current frame background image refers to the background image corresponding to the current original frame image output by the infrared detector. The background image refers to the output image of the infrared detector when the infrared detector is placed in the radiation field of the blackbody radiation source, that is, the blackbody background image. Correcting the current frame original image output by the infrared detector based on the current background image is to correct the non-uniformity of the current frame original image output by the infrared detector based on the calibration method. Using the current frame background image as a non-uniformity noise template in the current frame original image can eliminate most of the non-uniformity caused by temperature drift in the current frame original image.

[0035] In some embodiments, S02 specifically comprises: obtaining a current frame background image corresponding to the infrared detector at a current working temperature according to a fitting function between the background image and the working temperature and the current working temperature of the infrared detector, and correcting the current frame original image by using the current frame background image to obtain a current frame calibration corrected image. The fitting function is a function obtained by polynomial fitting of a plurality of background images corresponding to the infrared detector at a plurality of different working temperatures, the plurality of working temperatures are respectively temperatures in a normal working temperature range of the infrared detector, and the number of polynomial fitting is greater than or equal to 4.

[0036] The fitting function between the background image and the working temperature represents the relationship between the output image of the infrared detector and the FPA (Focal Plane Array) temperature when the infrared detector is placed in the radiation field of the blackbody radiation source and the working temperature of the infrared detector is in the normal working temperature range.

[0037] The fitting function between the background image and the working temperature can be obtained by calibrating the focal plane array of the infrared detector by the uniform radiation generated by the blackbody at a specific working temperature, and the obtained fitting function is stored. Specifically, in some embodiments, the step of obtaining the fitting function comprises: determining the normal working temperature range of the infrared detector, controlling the working temperature of the infrared detector to change in the normal working range by using a temperature box, collecting uniform background images at different working temperatures after the infrared device containing the infrared detector is stabilized, and performing function fitting with the working temperature T as the independent variable and the background gray value P(i,j) of each pixel point in the uniform background image as the dependent variable to obtain the fitting function between the background gray value P(i,j) of each pixel point and the working temperature of the infrared detector. The fitting function between the background image and the working temperature includes the fitting function between the background gray value P(i,j) of each pixel point and the working temperature T. The uniform background image and the background image, in the present application, the working temperature of the infrared detector refers to the FPA temperature of the infrared detector. The fitting function includes the fitting function corresponding to different pixel points in the background image. The current working temperature refers to the working temperature corresponding to the output of the current original image by the infrared detector.

[0038] Specifically, in some embodiments, the fitting function between the background image and the working temperature is a function obtained by fitting a plurality of background images corresponding to a plurality of different working temperatures of the infrared detector according to a polynomial fitting with a fitting degree greater than or equal to 4. The plurality of working temperatures are working temperatures in a normal working temperature range of the infrared detector. Taking the fitting degree of 4 as an example, four uniform background images at different working temperatures are collected, and a function fitting is performed on the background gray value P(i, j) of each pixel point in the uniform background image with the working temperature T as the independent variable to obtain the fitting function between the background gray value P(i, j) of each pixel point in the background image and the working temperature of the infrared detector, as shown in the following function (1):

[0039] P(i, j) = a4*T^4 + a3*T^3 + a2*T^2 + a1*T + a0 (1)

[0040] The fitting function between the background gray value P(i, j) of each pixel point and the working temperature T has five groups of coefficients (a0, a1, a2, a3, and a4), and the five groups of coefficients can be obtained according to the background gray values of the corresponding pixel points in the four uniform background images at different working temperatures, so as to determine and store the fitting function between the background gray value of each pixel point and the working temperature. Compared with linear fitting, the polynomial fitting with a suitable fitting degree can more accurately represent the actual relationship between the background image and the working temperature, and the computing power requirement is not high.

[0041] The current frame background image is a background image obtained by substituting the current working temperature of the infrared detector into the fitting function between the background image and the working temperature. The correction of the current frame original image using the current frame background image means that the background gray value of each pixel point in the current frame background image is used to correct the gray value of each pixel point in the current frame original image, so as to obtain the gray value of each pixel point after correction, thereby obtaining the current frame calibration and correction image. For example, the current working temperature of the infrared detector is T A , the background gray value of each pixel point in the current frame background image is P A (i, j), the original gray value of each pixel point in the current frame original image of the infrared detector is I A (i, j), and the average value of the background gray value of each pixel point in the current frame background image is avrI A , then the calibration and correction gray value I A calib(i, j) of each pixel point in the current frame calibration and correction image corresponds to the correction formula as shown in the following formula (2):

[0042] I A calib(i, j) = I A(i,j)-P A (i,j)+avrI A (2)

[0043] Therefore, in some embodiments, after determining the current frame background image of the infrared detector, the calibration and correction gray scale value of each pixel point is obtained according to the background gray scale value of each pixel point in the current frame background image, the original gray scale value of each pixel point in the current frame original image, the average value of the background gray scale value of each pixel point in the current frame background image, and the above formula (2), so as to obtain the pre-frame calibration and correction image after calibration and correction.

[0044] S04: judging whether the target scene currently detected by the infrared detector is in a motion state.

[0045] The target scene can be directly determined based on the original image output by the infrared detector whether it is in a motion state, or whether the infrared device containing the infrared detector is currently in a motion state can be sensed by a corresponding sensor. If the infrared device is in a motion state, the target scene is also in a motion state. Conversely, if the infrared device is in a stationary state, the target scene is also in a stationary state.

[0046] S06: if the target scene is currently in a motion state, correcting the current frame calibration and correction image based on the current frame scene correction matrix to obtain a current frame scene correction image, and performing image processing on the current frame scene correction image to obtain a target infrared image.

[0047] The current frame scene correction matrix is a scene correction matrix obtained by weighting the non-uniform noise matrix in the current frame calibration and correction image and the previous frame scene correction matrix. Each frame of the original image output by the infrared detector has a corresponding scene correction matrix. The previous frame scene correction matrix is the scene correction matrix corresponding to the previous frame original image.

[0048] If the non-uniform noise matrix of the current frame calibration and correction image is represented by imgHn(i,j), the corresponding weight coefficient is the first weight coefficient K1, the weight coefficient corresponding to the previous frame scene correction matrix imgHn-1(i,j) is 1-K1, and the acquisition formula of the current frame scene correction matrix imgN(i,j) is shown in the following formula (3):

[0049] imgN(i,j)=K1*imgHn(i,j)+(1-K1)*imgHn-1(i,j) (3)

[0050] Wherein, the value range of K1 is [0, 1]. The previous frame scene correction matrix imgHn-1(i,j) is the result obtained by iteration based on formula (3) in the previous frame.

[0051] In some embodiments, the current frame scene correction matrix imgN(i,j) can be used to correct the current frame calibration correction image imgn(i,j) according to the following formula (4) to obtain a current frame scene correction image imgn'(i,j):

[0052] imgn'(i,j) = imgn(i,j) - imgN(i,j) (4)

[0053] In some embodiments, the image processing algorithm is specifically an image noise reduction algorithm.

[0054] S08: If the target scene is currently in a static state, the current frame calibration correction image is processed by an image processing algorithm to obtain a target infrared image.

[0055] In the current target scene is in a static state, only based on the current frame background image, the current frame original image is corrected by non-uniformity calibration, and the current frame calibration correction image after non-uniformity calibration is used as the image to be processed by the image processing algorithm in the back end. In the current target scene is in a static state, first, based on the current frame background image, the current frame original image is corrected by non-uniformity calibration to obtain the current frame original image after calibration correction, i.e., the current frame calibration correction image, to eliminate most non-uniformity caused by temperature drift in the current frame original image. Then, based on the scene correction matrix, the current frame original image after calibration correction is corrected based on the scene to obtain the current frame scene correction image. The current frame scene correction image is the image obtained by correcting the current frame original image by calibration correction and scene-based correction, which is used as the image to be processed by the image processing algorithm in the back end.

[0056] In summary, the image processing method provided by some embodiments of the present application does not need to use a shutter to correct the non-uniformity of the original image output by the infrared detector, which can reduce the product size and weight of the infrared device in which the infrared detector is located, and effectively saves the manufacturing cost. In addition, the image processing method provided by some embodiments of the present application, in the current target scene is in a static state, combines the calibration-based correction method and the scene-based correction method to correct the non-uniformity of the original image output by the infrared detector, which effectively improves the image quality of the target infrared image. For details, please refer to Figure 2 and Figure 3 wherein Figure 2 is the original infrared image without non-uniformity correction, and Figure 3 is the target infrared image obtained by the non-uniformity correction processing based on the image processing method provided by some embodiments of the present application. Obviously, Figure 3 the image quality of the target infrared image in is significantly higher thanFigure 2 the original infrared image in the step S03.

[0057] Referring to Figure 4 As shown in FIG. 4, in the image processing method provided by some embodiments of the present application, the frame difference method is used to determine whether the target scene is currently in a motion state. Specifically, the above S04 specifically includes S041, S043 and S045, and the specific description of each step is as follows.

[0058] S041: determining whether the number of motion pixel points in the current frame of calibrated correction image meets a first preset condition, and the difference between the motion pixel point in the current frame of calibrated correction image and the corresponding pixel point in the previous frame of calibrated correction image meets a second preset condition.

[0059] The previous frame of calibrated correction image refers to the previous frame of calibrated correction image corresponding to the previous frame of original image adjacent to the current frame of original image. In some embodiments, the first preset condition is that the number of pixel points in S041 is greater than or equal to a preset number Nt, and in other embodiments, the first preset condition can also be that the number of pixel points in S041 is within a preset threshold number range. In some embodiments, the second preset condition is that the difference in S041 is greater than or equal to a first preset threshold Tn. In other embodiments, the second preset condition can also be that the difference in S041 is within a second preset threshold range. After the difference (subtraction) between the current frame of calibrated correction image and the previous frame of calibrated correction image is obtained, the difference of each pixel point is obtained, the greater the intensity of the motion of the target scene, the greater the difference, and the greater the number of motion pixel points. Therefore, the first preset threshold Tn and the preset number Nt can be correspondingly graded, respectively, to obtain nt level threshold parameters, where nt is greater than or equal to 3. The higher the level number (nt) corresponding to the threshold parameters reached by the difference and the number of pixels in S041, the more intense the current motion of the target scene.

[0060] When the result of the determination in S042 is yes, S043 is executed, otherwise S045 is executed.

[0061] S043: determining that the target scene is currently in a motion state.

[0062] S045: determining that the target scene is currently in a static state.

[0063] In the image processing method provided by some embodiments of the present application, an algorithm is used to determine whether the target scene is currently in a motion state, without using auxiliary devices such as gyroscopes, which is conducive to cost saving.

[0064] In some embodiments, step S06 specifically includes: using a multi-scale high-pass filtering method to exclude scene information in the current frame calibration and correction image, obtaining a non-uniform noise matrix of the current frame calibration and correction image; weighting the non-uniform noise matrix with the scene correction matrix of the previous frame to obtain the current frame scene correction matrix; and using the current frame scene correction matrix to correct the current frame calibration and correction image to obtain the current frame scene correction image. The specific process of weighting the current frame calibration and correction image with the non-uniform noise matrix to obtain the current frame scene correction matrix has been described above and will not be repeated here.

[0065] This embodiment focuses on describing the method for obtaining the non-uniform noise matrix of the current frame calibration image. After the high-frequency information in the current frame calibration image is extracted and calculated, it is necessary to filter the high-frequency information. Without filtering, severe ghosting will occur. Since strong textures in the scene are generally prone to causing ghosting, a multi-scale high-pass filtering method is used to exclude scene information in the current frame calibration image, obtaining the non-uniform noise matrix of the current frame calibration image. This helps to suppress ghosting in the target infrared image and improves the imaging quality of the target infrared image.

[0066] Please see Figure 5 As shown, in some embodiments, a multi-scale high-pass filtering method is used to exclude scene information in the current frame calibration and correction image to obtain the non-uniform noise matrix of the current frame calibration and correction image, specifically including S0611, S0613, S0615, S0617 and S0619, and the description of each step is as follows.

[0067] S0611: Perform low-pass filtering on the current frame calibration and correction image at the first scale to obtain the first low-frequency matrix.

[0068] S0613: Subtract the current frame calibration correction image from the first low-frequency matrix to obtain the first high-frequency matrix.

[0069] S0615: Perform a second-scale low-pass filter on the first high-frequency matrix to obtain the second low-frequency matrix.

[0070] S0617: Subtract the first high-frequency matrix from the second low-frequency matrix to obtain the second high-frequency matrix.

[0071] S0619: Subtract the first high-frequency matrix from the second high-frequency matrix to obtain the non-uniform noise matrix of the current frame calibration and correction image.

[0072] In some embodiments, the current frame calibration correction image is first low-pass filtered at a first scale, then the original image before the first scale low-pass filtering (the current frame calibration correction image) is subtracted from the low-pass filtered result to extract a first high-frequency matrix imgH1(i,j), then the first high-frequency matrix imgH1(i,j) is filtered according to certain conditions to filter out a non-uniform noise matrix imgHn(i,j). The specific filtering method is: the first scale low-pass filtering result is low-pass filtered again at a second scale, then the original image before the second filtering (the first high-frequency matrix) is subtracted from the second filtering result to extract a second high-frequency matrix imgH2(i,j), and the non-uniform noise matrix imgHn(i,j) is obtained according to the first high-frequency matrix imgH1(i,j) and the second high-frequency matrix imgH2(i,j), and the calculation formula of the non-uniform noise matrix imgHn(i,j) is shown in the following formula (5):

[0073] imgHn(i,j) = imgH1(i,j) - imgH2(i,j) (5)

[0074] Please refer to Figure 6 In some embodiments, the non-uniform noise matrix in the current frame calibration correction image is weighted with the scene correction matrix of the previous frame calibration correction image to obtain the scene correction matrix of the current frame calibration correction image, and the specific steps include S0621, S0623 and S0625, and each step is described as follows.

[0075] S0621: Determine the current motion level of the target scene.

[0076] The determination of the current motion level of the target scene can refer to the method described above, that is, the above level nt is the current motion level of the target scene. Here, other specific implementation methods for determining the current motion level of the target scene are not listed.

[0077] S0623: According to the current motion level, determine the first weight coefficient corresponding to the non-uniform noise matrix, and the higher the current motion level, the more intense the current motion of the target scene, and the smaller the first weight coefficient.

[0078] The larger nt is, the higher the current motion level of the target scene, which means that the current motion of the target scene is more intense, and the faster the iteration speed of the adaptive adjustment of the iteration speed of the infrared non-uniformity correction according to the motion level of the target scene is set, and the smaller the first weight coefficient. The first weight coefficient in S0623 is K1 in formula (3).

[0079] S0625: According to the first weight coefficient, the non-uniform noise matrix is weighted with the scene correction matrix of the previous frame to obtain the scene correction matrix of the current frame.

[0080] In some embodiments, the image processing method provided in the present application further comprises performing burn correction on the infrared image. Please refer to Figure 7 As shown in the figure, the burn correction on the infrared image specifically comprises S092, S094 and S096, and the description of each step is as follows.

[0081] S092: determining whether there is a current frame burn point in the current frame original image.

[0082] When the infrared detector is at high temperature, there may be some pixels in the focal plane array that are burned. The pixel points corresponding to these burned pixels in the image are called burn points. The current frame burn point refers to the burn point in the current frame original image.

[0083] S094: when there is a current frame burn point in the current frame original image and the target scene is currently in a motion state, determining a current frame burn correction matrix.

[0084] The burn correction matrix is a correction parameter for burn repair of the burn point in the current frame original image. After superimposing the burn correction matrix and the non-uniformity correction matrix, the original image is corrected, which can reduce the adverse effects on the image quality caused by the burn of the focal plane array. The non-uniformity correction matrix here can be a matrix corresponding to the background image representing the calibration correction of the original image, or a scene correction matrix.

[0085] S096: using the next frame background image superimposed with the current frame burn correction matrix to correct the next frame original image to obtain a next frame calibration correction image.

[0086] The next frame background image superimposed with the current frame burn correction matrix is to superimpose the background gray value of each pixel point in the next frame background image and the burn correction value of the burn point corresponding to the current frame burn matrix value. If the background gray value of each pixel point in the next frame background image is PB(i,j), the burn correction value of the burn point corresponding to the current frame burn matrix value is ΔA(i,j), the original gray value of each pixel point in the next frame original image of the infrared detector is IB(i,j), and the average value of the background gray value of each pixel point in the next frame background image is avrIB, then in some embodiments, the calculation formula of the calibration correction gray value IBcalib(i,j) of each pixel point in the next frame calibration correction image is shown in the following formula (5):

[0087] IBcalib(i,j) = IB(i,j) - [PB(i,j) + ΔA(i,j)] + avrIB (5)

[0088] Please refer to Figure 8In some embodiments, the specific steps of determining the current frame burn correction matrix in S094 include S0941, S0943 and S0945, and the descriptions of the respective steps are as follows.

[0089] S0941: Inter-frame registration is performed on the current frame original image and the previous frame original image to obtain the relative displacement between the current frame original image and the previous frame original image.

[0090] S0943: According to the first pixel coordinates of each current frame burn point in the current frame original image and the relative displacement, the second pixel coordinates of the corresponding points of each current frame burn point in the previous frame original image are determined.

[0091] S0945: The pixel difference between the first pixel coordinates of each current frame burn point and the corresponding second pixel coordinates is respectively weighted with the corresponding element in the previous frame burn correction matrix to obtain the current frame burn correction matrix.

[0092] Wherein, the higher the current motion level of the target scene is, the smaller the second weight coefficient corresponding to each pixel difference is when each pixel difference is weighted with the corresponding element in S0945. Inter-frame registration is performed on the current frame original image and the previous frame original image to determine the relative displacement dx, dy between the current frame original image and the previous frame original image. Let the corresponding coordinates of the current frame burn point in the current frame original image be the first pixel coordinates P1(i, j), then the coordinates of the corresponding points of the current frame burn point in the previous frame original image are the second pixel coordinates P2(i+dx, j+dy). If P1(i, j) ∈ S, S is a burn point set, then the calculation formula of the difference matrix Δn(i, j) composed of each pixel difference in S0945 is shown in the following formula (6):

[0093] Δn(i, j) = P(i, j) - P1(i+dx, j+dy) (6)

[0094] The calculation formula of the current frame burn correction matrix ΔN(i, j) is shown in the following formula (7):

[0095] ΔN(i, j) = K2*Δn(i, j) + (1-K2)*ΔN-1(i, j) (7)

[0096] Wherein, ΔN-1(i,j) in the formula (7) is the previous frame burn correction matrix. When S0945 is performed, K2 is the second weight coefficient corresponding to each differential pixel of Δn(i,j) in the pixel differential matrix, and (1-K2) is the weight coefficient corresponding to each element of the previous frame burn correction matrix. The adjustment manner of the second weight coefficient K2 is the same as that of the first weight coefficient, that is, the greater the current motion level nt of the target scene, the smaller K2. In some embodiments, the second weight coefficient K2 is the same weight coefficient as the first weight coefficient K1.

[0097] In some embodiments, the current frame scene correction image is obtained by correcting the current frame calibration correction image based on the current frame scene correction matrix in S006, which is performed after the above S094. For details, please refer to Figure 9 As shown in the figure, the current frame scene correction image is obtained by correcting the current frame calibration correction image based on the current frame scene correction matrix in S006, which includes S0631 and S0633, which are described as follows.

[0098] S0631: Superimpose the current frame scene correction matrix and the current frame burn correction matrix to obtain a current frame burn scene correction matrix.

[0099] S0633: Subtract the current frame calibration correction image from the current frame burn scene correction matrix to obtain the current frame scene correction image.

[0100] If the current frame scene correction matrix is imgN(i,j), the current frame calibration correction image is imgn(i,j), the current frame burn correction matrix is ΔN(i,j), and the current frame scene correction image imgn’(i,j) is obtained by S0633, the calculation formula of the current frame scene correction image imgn’(i,j) is shown in formula (8):

[0101] imgn’(i,j)=imgn(i,j)-[imgN(i,j)+ΔN(i,j)] (8)

[0102] In some embodiments, please refer to Figure 10 As shown in the figure, S092 specifically includes S0921, S0923 and S0925, and the description of each step is as follows.

[0103] The image processing method provided by some embodiments of the present application superimposes a burn correction matrix in the corresponding correction matrix for burn correction when correcting the non-uniformity of the infrared image, without the need for a shutter or an attenuation sheet for burn repair, which is conducive to cost saving.

[0104] S0921: Determine whether there is a burn risk point with a gray value meeting the burn risk condition in the current frame original image.

[0105] In some embodiments, the burn risk condition can refer to the gray value being greater than or equal to a second preset threshold value Tv. In other embodiments, the burn risk condition can also refer to the gray value being within a second preset threshold value range.

[0106] The burn risk point refers to a pixel point in the current frame original image that can have a burn risk. After determining that the current frame original image has a burn risk point, it is necessary to further determine whether there is a burn point in these burn risk points, otherwise it is determined that the current frame original image does not have a current frame burn point. S0923: If there is a burn risk point in the current frame original image, and the proportion of the saturated point in the burn risk point meets the burn condition, it is determined that the current frame original image has a corresponding current frame burn point.

[0107] In some embodiments, the burn condition can refer to the proportion in S0923 being greater than or equal to a third preset threshold value Pr. In other embodiments, the burn condition can also refer to the proportion in S0923 being within a third preset threshold value range. The saturated point refers to a pixel point whose gray value reaches the upper limit. If the saturated point meets the fourth preset condition, it means that the saturated point is burned, and the saturated point in the burn risk point at this time is the current frame burn point.

[0108] S0925: Each saturated point is determined as a current frame burn point, and each current frame burn point is updated to the burn point set.

[0109] The burn point set is S = {P1(i, j) | P1(i, j) is the original gray value corresponding to the current frame burn point}. Since the infrared detector is slow to recover after being burned, it is necessary to correct the burn point in advance when the next time the machine is started, that is, the difference value Δ of the burn point obtained in the last iteration in the previous start-up process needs to be stored and added to the background gray value P of the corresponding pixel point in the first iteration of the next start-up, and the burn point set S is empty until the next time the burn is detected.

[0110] Specifically, in some embodiments, if the current frame original image is the first frame original image of the current start-up of the infrared detector, the correction of the current frame original image based on the current frame background image in S02 above to obtain the current frame calibration correction image specifically includes: using the current frame background image with the burn correction matrix of the last frame in the previous start-up process of the infrared detector to correct the current frame original image to obtain the current frame calibration correction image, and clearing the burn point set.

[0111] In some embodiments, S08 specifically comprises: judging whether the variation amplitude of the current working temperature of the infrared detector relative to the state switching temperature meets a third preset condition, and if so, performing adaptive noise reduction processing on each current frame calibration correction image based on texture information representing the texture strength of each region in the current frame calibration correction image, the weaker the texture of a region, the greater the corresponding noise reduction strength.

[0112] The state switching temperature is the working temperature of the infrared detector when the target scene switches from a motion state to a static state. The third preset condition can refer to the variation amplitude being greater than or equal to a fourth preset threshold, and the fourth preset threshold can be, for example, the variation amplitude being greater than or equal to 3. Since non-uniform noise is strongly related to FPA temperature, if the FPA temperature does not change much, the non-uniform noise will also not be heavy, in which case there is no need to enable the adaptive noise reduction mode of the backend, otherwise, the backend noise reduction mode needs to be enabled, that is, adaptive noise reduction processing is performed on each current frame calibration correction image based on texture information representing the texture strength of each region in the current frame calibration correction image. Therefore, the adaptive noise reduction mode of the backend can be determined according to the variation of the FPA temperature.

[0113] When the target scene is currently in a static state, it is not possible to correct the current frame calibration correction image again based on the scene-based non-uniform correction method. In order to better improve the image quality of the target infrared image, in some embodiments, the influence of non-uniformity when the target scene is in a static state is reduced by combining the FPA temperature of the infrared detector and the noise reduction strength in the enhanced backend processing algorithm. Specifically, in order to minimize the loss of details, only the regions with less texture details in the current frame calibration correction image are increased in noise reduction strength of the backend image processing algorithm.

[0114] Further, the adaptive noise reduction processing of each region based on the texture information representing the texture strength of each region in the current frame calibration correction image specifically comprises: calculating the variance of each pixel point in a preset neighborhood in the current frame calibration correction image, and performing adaptive noise reduction processing on the previous frame calibration correction image according to the variances, the smaller the variance of a preset neighborhood, the weaker the texture of the preset neighborhood, and the greater the corresponding noise reduction strength.

[0115] Specifically, the texture strength of each region in the current frame calibration correction image can be determined by calculating the variance in a 7*7 window, and the variance obtained here is the texture information described above. The smaller the variance, the greater the noise reduction strength.

[0116] Referring to FIG. 1, Figure 11 FIG. 1 shows a calibration correction flowchart in an image processing method according to some embodiments of the present application. The steps of calibration correction of the original image output by the infrared detector include S11, S12, S13, S14 and S15, and the descriptions of each step are as follows:

[0117] S11: Collecting blackbody background image according to FPA temperature. That is, collecting blackbody background image at different working temperatures in the normal working temperature range of the infrared detector, to obtain the corresponding background image of the infrared detector at each different working temperature.

[0118] S12: Polynomial fitting of the relationship between the background pixels of each pixel point and the FPA temperature. The relationship in S12 is the fitting function of each pixel point obtained above. The fitting functions of each pixel point constitute the fitting function between the background image of the infrared detector and the working temperature.

[0119] S13: Storing the polynomial coefficients corresponding to each pixel point. The polynomial coefficients in S13 are a0 to a4 above. Storing each polynomial coefficient, that is, storing the fitting function between different pixel points and working temperature.

[0120] S14: Reading FPA temperature during actual use, and calculating background data at the temperature according to the coefficients. Obtain the current FPA temperature of the infrared detector, and substitute the stored polynomial coefficients and the current FPA temperature into the fitting function corresponding to each pixel point to obtain the background data of each pixel point, that is, the background gray value of each pixel point, thereby obtaining the current frame background image of the infrared detector at the current FPA temperature.

[0121] S15: Superimposing the burn correction matrix for calibration correction. Specifically, superimposing the background gray value of each pixel point in the current frame background image with the corresponding element in the previous frame burn correction matrix, and then using the current frame background image superimposed with the previous frame burn correction matrix to correct the current frame original image to obtain the current frame calibration correction image.

[0122] Please refer to Figure 12 The image processing method provided by some embodiments of the present application is shown in the figure. In some embodiments, the image processing method includes S21, S22, S23, S24, S25, S26, S27, S28, S29, S210, S211, S212, and S213. The description of each step is as follows.

[0123] S21: Burn detection. That is, comparing the original gray value of each pixel point in the current frame original image with the corresponding threshold value respectively to detect whether there is a burn risk area in the current frame original image.

[0124] S22: Burn determination. Determine whether there is a burn point in the burn risk points determined in S21, that is, determine whether the saturated point in the burn risk point reaches the corresponding threshold value. If yes, it is determined that there is a current frame burn point in the current frame original image, and the result output by S22 is yes.

[0125] S23: whether motion. That is, judging whether the target scene is currently in a motion state. If yes, and the result of S22 is also yes, then performing S24 to correct the burn in the target scene in a motion state based on inter-frame registration. If the result of S23 is no, then S211 needs to be entered.

[0126] S24: inter-frame registration. The specific implementation of inter-frame registration of the current frame original image and the previous frame original image is described above and will not be repeated here.

[0127] S25: calculating the burn correction parameter. That is, calculating Δn(i,j) in the pixel difference matrix described above, which can be referred to the corresponding description above.

[0128] S26: determining the inter-frame burn correction iteration coefficient according to the motion intensity, and performing burn correction matrix iteration calculation according to the burn correction iteration coefficient. The motion intensity is the aforementioned current motion level nt, and the inter-frame burn correction iteration coefficient in S26 is the aforementioned second weight coefficient K2. According to the adaptive iteration adjustment setting based on the motion intensity, the larger the motion level nt, the smaller the second weight coefficient K2, and the faster the iteration speed, so as to obtain the current frame burn correction matrix ΔN(i,j).

[0129] S27: judging the motion intensity. That is, determining nt described above. S27 is executed after S23 and before S26.

[0130] S28: calculating the single-frame high frequency to extract the non-uniform noise. That is, determining the non-uniform noise matrix in the current frame calibration correction image.

[0131] S29: determining the inter-frame scene correction iteration coefficient according to the motion intensity, and performing scene correction matrix iteration calculation according to the scene correction iteration coefficient. The motion intensity is the aforementioned current motion level nt, and the inter-frame scene correction iteration coefficient in S29 is the aforementioned first weight coefficient K1. According to the adaptive iteration adjustment setting based on the motion intensity, the larger the motion level nt, the smaller the first weight coefficient K1, and the faster the iteration speed, so as to obtain the current frame scene correction matrix imgN(i,j).

[0132] S210: The result is taken as a correction matrix. That is, after superimposing the current frame scene correction matrix imgN(i,j) and the current frame burn correction matrix AN(i,j), the correction matrix imgN(i,j)+AN(i,j) is obtained, then the current frame calibration correction image is corrected by using the correction matrix imgN(i,j)+AN(i,j), the current frame scene correction image imgn'(i,j)=imgn(i,j)-[imgN(i,j)+AN(i,j)] is obtained, and the current frame scene correction image imgn'(i,j) is output as a corrected result to be processed by the image processing algorithm of the back end.

[0133] S211: Whether the FPA temperature change exceeds a threshold. When the judgment in S23 is no, S211 is entered, that is, when the target scene is currently in a static state, it is necessary to determine whether the adaptive noise reduction mode needs to be started according to the change amplitude of the FPA temperature. If the FPA temperature change exceeds the threshold, S212 is entered.

[0134] S212: Calculate the variance in the neighborhood of the pixel point. The variance in the neighborhood is taken as the texture information representing the texture strength in the neighborhood.

[0135] S213: Different intensity noise reduction is performed according to the variance. The smaller the variance, the greater the noise reduction intensity of the corresponding neighborhood.

[0136] From the above, the image processing method provided by some embodiments of the present application can at least achieve one of the following beneficial effects:

[0137] 1. Non-uniformity correction does not need to use a shutter, which can reduce product size and weight, and save cost.

[0138] 2. No auxiliary equipment such as a gyroscope is needed to determine the motion state of the target scene, which saves cost.

[0139] 3. The ghost suppression strategy of extracting a non-uniformity noise matrix by using a multi-scale high-pass filter can greatly suppress the ghost generated in the scene-based non-uniformity correction process.

[0140] 4. When the target scene is in a static state, adaptive noise reduction is performed in combination with the FPA temperature and the image processing algorithm of the back end, which ensures the image effect in the static state.

[0141] 5. Burn repair does not need to use a shutter, and has a calibration data correction mechanism.

[0142] Please refer to Figure 13As shown, it is a structure schematic diagram of the infrared image processing device based on shutterless non-uniformity correction provided by some embodiments of the present application. In some embodiments, the infrared image processing device provided by the present application comprises a calibration correction module 101, a state judgment module 102, a scene correction processing module 103 and a calibration correction processing module 104. The calibration correction module 101 is specifically configured to correct the current frame original image output by the infrared detector based on the current frame background image to obtain a current frame calibration correction image. The state judgment module 102 is specifically configured to judge whether the target scene currently detected by the infrared detector is in a motion state. The scene correction processing module 103 is specifically configured to correct the current frame calibration correction image based on the current frame scene correction matrix when the target scene is currently in a motion state, to obtain a current frame scene correction image, and to perform image processing on the current frame scene correction image to obtain a target infrared image. The calibration correction processing module 104 is specifically configured to perform image processing on the current frame calibration correction image to obtain a target infrared image when the target scene is currently in a stationary state. The image processing device provided by the embodiments of the present application can achieve the same technical effects as the image processing method provided by the embodiments of the present application, and will not be repeated here.

[0143] Optionally, the calibration correction module 101 is also specifically configured to obtain the current frame background image corresponding to the current working temperature of the infrared detector according to the fitting function between the background image and the working temperature and the current working temperature of the infrared detector, and correct the current frame original image using the current frame background image to obtain the current frame calibration correction image. The fitting function is a function obtained by polynomial fitting of a plurality of background images corresponding to a plurality of different working temperatures of the infrared detector, the plurality of working temperatures are respectively temperatures in the normal working temperature range of the infrared detector, and the number of polynomial fitting is greater than or equal to 4.

[0144] The state judgment module 102 is also specifically configured to judge whether the number of motion pixel points in the current frame calibration correction image satisfies a first preset condition, and whether the difference between the motion pixel points in the current calibration correction image and the corresponding pixel points in the previous frame calibration correction image satisfies a second preset condition, if yes, it is judged that the target scene is currently in a motion state, if not, it is judged that the target scene is currently in a stationary state. The scene correction processing module 103 is also specifically configured to exclude the scene information in the current frame calibration correction image by using a multi-scale high-pass filtering method to obtain a non-uniformity noise matrix of the current frame calibration correction image, and to weight the non-uniformity noise matrix and the previous frame scene correction matrix to obtain the current frame scene correction matrix, and to correct the current frame calibration correction image using the current frame scene correction matrix to obtain the current frame scene correction image.

[0145] Optionally, the scene correction processing module 103 is further configured to: perform low-pass filtering on the current frame of the calibration corrected image at a first scale to obtain a first low-frequency matrix; subtract the first low-frequency matrix from the current frame of the calibration corrected image to obtain a first high-frequency matrix; perform low-pass filtering on the first high-frequency matrix at a second scale to obtain a second low-frequency matrix; subtract the second low-frequency matrix from the first high-frequency matrix to obtain a second high-frequency matrix; and subtract the second high-frequency matrix from the first high-frequency matrix to obtain a non-uniformity noise matrix of the current frame of the calibration corrected image.

[0146] The scene correction processing module 103 is further configured to: determine a current motion level of the target scene.

[0147] According to the current motion level, a first weight coefficient corresponding to the non-uniformity noise matrix is determined, and the higher the current motion level, the more intense the current motion of the target scene, and the smaller the first weight coefficient; and according to the first weight coefficient, the non-uniformity noise matrix is weighted with a previous frame of the scene correction matrix to obtain a current frame of the scene correction matrix.

[0148] Optionally, the image processing apparatus provided by some embodiments of the present application further comprises a burn repair correction module. The burn repair correction module is specifically configured to: determine whether there is a current frame of burn point in a current frame of original image; when there is a current frame of burn point in the current frame of original image and the target scene is currently in a motion state, determine a current frame of burn correction matrix; and correct a next frame of original image by using a next frame of background image superimposed with the current frame of burn correction matrix to obtain a next frame of calibration corrected image.

[0149] Optionally, the burn repair correction module is further configured to: perform inter-frame registration on the current frame of original image and a previous frame of original image to obtain a relative displacement between the current frame of original image and the previous frame of original image; according to first pixel coordinates of each current frame of burn point in the current frame of original image and the relative displacement, determine second pixel coordinates of corresponding points of each current frame of burn point in the previous frame of original image; and weight pixel difference values between the first pixel coordinates of each current frame of burn point and the corresponding second pixel coordinates with corresponding elements in a previous frame of burn correction matrix respectively to obtain the current frame of burn correction matrix. Wherein, the higher the current motion level of the target scene, the smaller the second weight coefficient corresponding to each pixel difference value when each pixel difference value is weighted with the corresponding element.

[0150] Optionally, the burn repair correction module is further configured to: determine a current motion level of the target scene; determine a second weight coefficient corresponding to the pixel difference according to the current motion level, the higher the current motion level, the more intense the current motion of the target scene, and the smaller the second weight coefficient; and weight the pixel difference between the first pixel coordinate and the corresponding second pixel coordinate of each burn point of the current frame according to the second weight coefficient, to obtain the burn correction matrix of the current frame.

[0151] Optionally, the burn repair correction module is further configured to: superimpose the current frame scene correction matrix and the burn correction matrix of the current frame; and subtract the current frame scene correction matrix after superimposing the burn correction matrix of the current frame from the calibrated image of the current frame to obtain the current frame scene correction image.

[0152] Optionally, the burn repair correction module is further configured to: determine whether there is a burn risk point with a gray value satisfying a burn risk condition in the current frame original image; if there is a burn risk point in the current frame original image and the proportion of the saturated points in the burn risk point satisfies the burn condition, determine that there is a corresponding current frame burn point in the current frame original image; determine each saturated point as the current frame burn point, and update each current frame burn point to the burn point set.

[0153] Optionally, the calibration correction module 101 is further configured to: if the current frame original image is the first frame original image in the current start-up process of the infrared detector, correct the current frame original image using the current frame background image superimposed with the burn correction matrix of the last frame in the previous start-up process of the infrared detector to obtain the calibrated image of the current frame, and clear the burn point set.

[0154] Optionally, the calibration correction processing module 104 is further configured to: determine whether the change range of the current working temperature of the infrared detector relative to the state switching temperature satisfies a third preset condition; if yes, perform adaptive noise reduction processing on each previous frame calibrated image based on the texture information representing the texture strength of each region in the current frame calibrated image, the weaker the texture of a region, the greater the corresponding noise reduction strength; wherein the state switching temperature is the working temperature of the infrared detector when the target scene switches from a motion state to a stationary state.

[0155] Optionally, the calibration correction processing module 104 is further configured to: calculate the variance of each pixel point in the preset neighborhood in the current frame calibrated image; and perform adaptive noise reduction processing on the previous frame calibrated image according to the variances, the smaller the variance of a preset neighborhood, the weaker the texture of the preset neighborhood, and the greater the corresponding noise reduction strength.

[0156] Please refer to Figure 14As shown in FIG. 1, it is a schematic diagram of an infrared device structure according to some embodiments of the present application. The infrared device provided by the embodiments of the present application comprises a memory 201 and a processor 202, the memory 201 stores a computer program executable by the processor 202, and the computer program is executed by the processor 202 to implement the image processing method according to any of the embodiments of the present application. The infrared device provided by some embodiments of the present application and the image processing method provided by some embodiments of the present application can achieve the same technical effects, which will not be repeated here.

[0157] In some embodiments, the infrared device provided by some embodiments of the present application further comprises a lens module 203 and an infrared detector 204. The lens module 203 is used to irradiate the infrared radiation of the target scene to the infrared focal plane array of the infrared detector 204. The infrared detector 204 is used to convert the infrared radiation into an infrared electrical signal, i.e. an original image, and input the original image to the processor 202 for processing to obtain a target infrared image.

[0158] Please refer to Figure 15 As shown in FIG. 2, it is a schematic diagram of the functional modules of the infrared device according to some embodiments of the present application. The functional modules of the infrared device provided by the embodiments of the present application comprise a lens module 301, an infrared detector module 302, a non-uniformity correction module 303, a burn detection and repair module 304, an image processing module 305 and a signal output module 306. Among them, the lens module 301 is used to collect infrared radiation; the infrared detector module 302 is used to convert the infrared radiation into an original infrared image output; the non-uniformity correction module 303 is used to correct the non-uniformity of the original infrared image; the burn detection and repair module 304 is used to detect and repair the original infrared image; the image processing module 305 is used to process the original infrared image after non-uniformity correction and burn repair correction to obtain a target infrared image; the signal output module 306 is used to output the target infrared image.

[0159] In some embodiments, the functions of the lens module 301 are realized by the lens module 203, the functions of the infrared detector module 302 are realized by the infrared detector 204, and the corresponding functions of the non-uniformity correction module 303, the burn repair module 303, the image processing module 305 and the signal output module 306 are realized by the processor 202.

[0160] The embodiment of the present application further provides a computer readable storage medium, and the computer readable storage medium stores a computer program. The computer program is executed by a processor to realize each process in the image processing method according to any one of the embodiments of the present application, and the same technical effects can be achieved. To avoid repetition, details are not described here. The computer readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0161] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An image processing method based on shutterless non-uniformity correction, characterized in that, include: The original image of the current frame output by the infrared detector is corrected based on the background image of the current frame to obtain the calibration correction image of the current frame; Determine whether the target scene detected by the infrared detector is currently in motion; If the target scene is currently in motion, the current frame calibration correction image is corrected based on the current frame scene correction matrix to obtain the current frame scene correction image, and the current frame scene correction image is processed to obtain the target infrared image. The step of correcting the current frame calibration correction image based on the current frame scene correction matrix to obtain the current frame scene correction image includes: using a multi-scale high-pass filtering method to remove scene information in the current frame calibration correction image to obtain the non-uniform noise matrix of the current frame calibration correction image. The non-uniform noise matrix is ​​weighted with the scene correction matrix of the previous frame to obtain the scene correction matrix of the current frame; the current frame calibration correction image is corrected using the current frame scene correction matrix to obtain the scene correction image of the current frame. If the target scene is currently stationary, then image processing is performed on the current frame calibration and correction image to obtain the target infrared image.

2. The image processing method according to claim 1, characterized in that, The process of correcting the original image of the current frame output by the infrared detector based on the background image of the current frame to obtain the calibration correction image of the current frame includes: Based on the fitting function between the background image and the operating temperature and the current operating temperature of the infrared detector, the current frame background image of the infrared detector at the current operating temperature is obtained. The original image of the current frame is corrected using the background image of the current frame to obtain the calibration correction image of the current frame; The fitting function is a function obtained by polynomial fitting based on multiple background images corresponding to multiple different operating temperatures of the infrared detector. The multiple operating temperatures are temperatures within the normal operating temperature range of the infrared detector, and the degree of the polynomial fitting is greater than or equal to 4.

3. The image processing method according to claim 1, characterized in that, The step of determining whether the target scene detected by the infrared detector is currently in motion includes: Determine whether the number of moving pixels in the current frame calibration correction image meets the first preset condition, and the difference between the moving pixels in the current frame calibration correction image and the corresponding pixels in the previous frame calibration correction image meets the second preset condition; If so, then it is determined that the target scene is currently in motion; If not, then the target scene is determined to be in a static state.

4. The image processing method according to claim 1, characterized in that, The process of using a multi-scale high-pass filtering method to remove scene information from the current frame calibration and correction image, resulting in a non-uniform noise matrix for the current frame calibration and correction image, includes: The current frame calibration and correction image is subjected to low-pass filtering at a first scale to obtain a first low-frequency matrix; The first high-frequency matrix is ​​obtained by subtracting the current frame calibration and correction image from the first low-frequency matrix. The first high-frequency matrix is ​​subjected to a second-scale low-pass filter to obtain the second low-frequency matrix; The second high-frequency matrix is ​​obtained by subtracting the first high-frequency matrix from the second low-frequency matrix. The non-uniform noise matrix of the current frame calibration and correction image is obtained by subtracting the first high-frequency matrix from the second high-frequency matrix.

5. The image processing method according to claim 1, characterized in that, The step of weighting the non-uniform noise matrix with the scene correction matrix of the previous frame to obtain the scene correction matrix of the current frame includes: Determine the current motion level of the target scene; Based on the current motion level, a first weighting coefficient corresponding to the non-uniform noise matrix is ​​determined. The higher the current motion level, the more intense the current motion of the target scene, and the smaller the first weighting coefficient. Based on the first weighting coefficient, the non-uniform noise matrix is ​​weighted with the previous frame scene correction matrix to obtain the current frame scene correction matrix.

6. The image processing method according to claim 1, characterized in that, Also includes: Determine whether there is a burn point in the original image of the current frame; When the burn point of the current frame exists in the original image of the current frame and the target scene is currently in motion, determine the burn correction matrix of the current frame; The original image of the next frame is corrected by superimposing the burn correction matrix of the current frame onto the background image of the next frame, so as to obtain the calibration correction image of the next frame.

7. The image processing method according to claim 6, characterized in that, Determining the burn correction matrix for the current frame includes: The current frame original image and the previous frame original image are registered to obtain the relative displacement between the current frame original image and the previous frame original image. Based on the first pixel coordinates of each current frame burn point in the current frame original image and the relative displacement, determine the second pixel coordinates of the corresponding point of each current frame burn point in the previous frame original image; The pixel difference between the first pixel coordinate and the corresponding second pixel coordinate of each burn point in the current frame is weighted with the corresponding element in the burn correction matrix of the previous frame to obtain the burn correction matrix of the current frame. Wherein, the higher the current motion level of the target scene, the smaller the second weight coefficient corresponding to the pixel difference when each pixel difference is weighted with the corresponding element.

8. The image processing method according to claim 6, characterized in that, The step of correcting the current frame calibration correction image based on the current frame scene correction matrix to obtain the current frame scene correction image includes: The current frame scene correction matrix and the current frame burn correction matrix are superimposed to obtain the current frame burn scene correction matrix; The current frame scene correction image is obtained by subtracting the current frame calibration correction image from the current frame burn scene correction matrix.

9. The image processing method according to any one of claims 1 to 8, characterized in that, If the target scene is currently stationary, then image processing is performed on the current frame calibration and correction image to obtain a target infrared image, including: When the target scene is currently in a static state, determine whether the change in the current operating temperature of the infrared detector relative to the state switching temperature meets the third preset condition; If so, based on the texture information used to characterize the texture strength of each region in the current frame calibration and correction image, adaptive noise reduction processing is performed on each current frame calibration and correction image. The weaker the texture of the region, the greater the corresponding noise reduction intensity. The state switching temperature is the operating temperature of the infrared detector when the target scene switches from a moving state to a stationary state.

10. An infrared device, comprising a memory and a processor, wherein the memory stores a computer program executable by the processor, and the computer program, when executed by the processor, implements the image processing method as described in any one of claims 1 to 9.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the image processing method as described in any one of claims 1 to 9.

Images