A method and device for processing consistency of dawn-dusk orbit satellite multi-camera low-light imaging

Abstract

The application discloses a method and equipment for consistency processing of dawn-dusk orbit satellite multi-camera low-light imaging, and relates to the technical field of remote sensing image processing. The method comprises the following steps: acquiring low-light imaging observation images and dawn-dusk orbit satellite observation information; correcting the differences between cameras in a multi-camera low-light imager for the low-light imaging observation images to obtain initial observation images; determining radiance data based on the observation information; adjusting the initial observation images based on the radiance data to obtain adjusted initial observation images; constructing a fitting curve of the solar zenith angle based on the observation information; correcting the adjusted initial observation images based on the fitting curve of the solar zenith angle to obtain corrected initial observation images; and performing uniformization processing on the corrected initial observation images to obtain corrected observation images. The application can obtain uniform and consistent corrected observation images.

Description

Technical Field

[0001] This application relates to the field of remote sensing image processing technology, and in particular to a consistency processing method and device for low-light imaging by multiple cameras on a dawn-dusk orbit satellite. Background Technology

[0002] The low-light imager onboard the dawn-dusk orbit satellite boasts a wide swath and large dynamic range, enabling it to acquire cloud information near the dawn-dusk transition zone as well as information on brighter city lights at night. However, due to the imager's dynamic detection range spanning seven orders of magnitude, it is difficult to simultaneously present both strong illumination (such as sunlight reflected from the Earth during the daytime) and weak illumination (such as moonlight reflected from the Earth) within a single grayscale image, posing a significant challenge to image display applications. Furthermore, when the payload response exhibits nonlinear decay in the low-end boundary region, significant differences appear between the observations of the high-brightness and low-brightness channels. This can result in "black patches" in the image due to weak signals or "white patches" due to over-correction.

[0003] For operational weather forecasting and short-term forecasting applications, image products are difficult for forecasters to understand without correction of the radiance acquired by the detectors. That is, images appear unusually bright on the sunlit side and relatively dark on the night side, requiring scaling transformations of the received radiance values ​​during data processing. Linear and logarithmic scaling transformations, two simple processing methods, are ineffective for imaging low-light channels and fail to represent regional environmental changes. In the inventors' known technology, a gain adjustment model based on the solar / lunar zenith angle is used in the production of Level 2 low-light product data and Level 1 NCC (Normalized Cross-Correlation) image products. While the physical meaning of this correction method is clear, it requires inputting an auxiliary optimization lookup table.

[0004] Another related technology is the HNCC (High Near-constant Contrast) scaling algorithm, applicable to the low-light channel imaging enhancement display of the SNPP satellite (Suomi National Polar-Orbiting Partnership) VIIRS DNB (Visible Infrared Imaging Radiometer Suite day / night band). This algorithm uses only the solar zenith angle, lunar zenith angle, and lunar phase angle information from the L1 data of the DNB's SDR (Science Data Record). It employs a fixed correction curve and a simple correction formula, eliminating the need for auxiliary data such as lookup tables, and can achieve low-light channel imaging with constant contrast under full illumination conditions. However, when directly applied to other satellite low-light payloads, especially in the region near the dawn-dusk transition zone, it is often affected by factors such as the nonlinearity of the payload performance response, making it difficult to obtain uniform and consistent low-light images. Summary of the Invention

[0005] The purpose of this application is to provide a consistency processing method and device for multi-camera low-light imaging of satellites in twilight orbit, which can obtain uniform and consistent corrected observation images.

[0006] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit, including: Acquire low-light imaging observation images and observation information from the dawn-dusk orbit satellite; the low-light imaging observation images are acquired by a multi-camera low-light imager mounted on the dawn-dusk orbit satellite; The low-light imaging observation image is subjected to difference correction between cameras in a multi-camera low-light imager to obtain the initial observation image; The radiance data is determined based on the observation information; A fitting curve for the solar zenith angle is constructed based on the radiance data; the initial observation image is corrected based on the fitting curve for the solar zenith angle to obtain a corrected initial observation image; The corrected initial observation image is then homogenized to obtain the corrected observation image.

[0007] In one embodiment, the low-light imaging observation image is subjected to difference correction between cameras in a multi-camera low-light imager to obtain an initial observation image, including: Based on the observation information, the average solar zenith angle of every two adjacent cameras in the multi-camera low-light imager is determined. Based on the observed average values ​​of the solar zenith angle within the first set angle range of the first camera and the second set angle range of the second camera in every two adjacent cameras, a quadratic nonlinear curve is obtained. Determine the difference between the observed mean of the solar zenith angle of the first camera and the quadratic nonlinear curve; Based on the difference value, the solar zenith angle of the two cameras is fitted twice to obtain the correction functions of the two cameras respectively; The low-light imaging observation image is corrected based on the correction function of each camera and a first set angle as a reference to obtain the initial observation image.

[0008] In one embodiment, the process of determining radiance data based on the observation information includes: Obtain laboratory calibration data for satellites in dawn-dusk orbit before their flight; Based on the laboratory calibration data, the observation information is converted into radiance values; The radiance value is logarithmically scaled to obtain the radiance data.

[0009] In one embodiment, the corrected initial observation image is homogenized to obtain a corrected observation image, including: A panoramic observation fit is performed on the corrected initial observation image to obtain a two-dimensional curved surface background function; Based on the two-dimensional curved background function, the ratio of each pixel to the background value in the corrected initial observation image is determined to obtain the homogenization processing result; Extract the enhanced display area from the homogenization result; The enhanced display range is subjected to a linear scaling transformation to obtain the corrected observation image.

[0010] In one embodiment, the process of constructing a fitting curve for the solar zenith angle based on the radiance data includes: Based on the radiance data, determine the radiance value of each camera in the multi-camera low-light imager at the solar zenith angle within the third set angle range, and determine the average radiance value of the multi-camera low-light imager at each angle within the third set angle range. A fitting curve for the solar zenith angle is constructed based on the average radiance values ​​at various angles.

[0011] In one embodiment, the initial observation image is corrected based on the fitted curve of the solar zenith angle to obtain a corrected initial observation image, including: The relative difference of the initial observation image with respect to the solar zenith angle at the second set angle is determined by using the fitted curve of the solar zenith angle. The initial observation image is corrected based on the relative difference to obtain the corrected initial observation image.

[0012] In one embodiment, the first set angle range is 85 degrees to 87 degrees.

[0013] In one embodiment, the second set angle range is 91 degrees to 93 degrees.

[0014] In one embodiment, the range of the enhanced display interval is [0.8, 1.2].

[0015] In a second aspect, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the consistency processing method for multi-camera low-light imaging of a twilight orbit satellite as described above.

[0016] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a consistency processing method and device for multi-camera low-light imaging of a twilight orbit satellite. It corrects the differences between cameras in the low-light imaging observation images, solving the problem of the impact of camera response differences on imaging consistency. Based on the observation information of the twilight satellite, it determines the radiance data, constructs a fitting curve based on the solar zenith angle, and sequentially performs radiance adjustment, fitting curve correction based on the solar zenith angle, and uniformization processing on the initial observation images to finally obtain a uniform and consistent corrected observation image. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a consistency processing method for multi-camera low-light imaging of a satellite in a dawn-dusk orbit, according to one embodiment of this application. Figure 2 This is a schematic diagram of the low-brightness and high-brightness channels observed in a certain sea area on a certain morning. Figure 3 This is a schematic diagram showing the variation of radiance of the observed images in the low-brightness and high-brightness channels with the solar zenith angle. Figure 4 This is a schematic diagram showing the observed images of the same area and the variation of the observed radiance values ​​with the solar zenith angle. Figure 5This is a schematic diagram of the processing flow for the low-brightness channel and the high-brightness channel; Figure 6 This is a schematic diagram showing the distribution of low-light imaging observation images and the corrected observation images obtained after correction processing as a function of the solar zenith angle. Figure 7 This is a schematic diagram of the frequency distribution of observation data for low-light imaging observation images and the corrected observation images obtained after correction processing. Figure 8 A schematic diagram illustrating the impact of the correction process on the image display effect; Figure 9 This is a comparison image of the stitched images after processing; Figure 10 This is a schematic diagram of the image after latitude and longitude projection following correction; Figure 11 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] In one exemplary embodiment, such as Figure 1 As shown, a consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit is provided, including: Step 1: Acquire low-light imaging observation images and observation information from the dawn-dusk orbit satellite. The low-light imaging observation images were acquired using a multi-camera low-light imager mounted on the dawn-dusk orbit satellite.

[0022] Step 2: Perform difference correction between cameras in the multi-camera low-light imager on the low-light imaging observation image to obtain the initial observation image.

[0023] Step 3: Determine radiance data based on observation information.

[0024] Step 4: Construct a fitting curve for the solar zenith angle based on the radiance data. Correct the initial observation image using the fitting curve for the solar zenith angle to obtain the corrected initial observation image.

[0025] Step 5: The corrected initial observation image is homogenized to obtain the corrected observation image.

[0026] As an optional implementation, in order to reduce the impact of inconsistencies between cameras in a multi-camera low-light imager on subsequent display, the implementation process of step 2 above includes: Step 21: Determine the average solar zenith angle of every two adjacent cameras in the multi-camera low-light imager based on the observation information.

[0027] Step 22: Based on the observed average values ​​of the solar zenith angle in the first set angle range of the first camera and the second set angle range of the second camera in every two adjacent cameras, a quadratic nonlinear curve is obtained.

[0028] Step 23: Determine the difference between the observed mean of the solar zenith angle of the first camera and the quadratic nonlinear curve.

[0029] Step 24: Perform a second fitting on the solar zenith angle of the two cameras based on the difference value to obtain the correction functions of the two cameras respectively.

[0030] Step 25: Based on the correction functions of each camera, the low-light imaging observation image is corrected using the first set angle as a reference to obtain the initial observation image.

[0031] The first set angle range is 85 degrees to 87 degrees. The second set angle range is 91 degrees to 93 degrees.

[0032] Step 3 above, the process of determining radiance data based on observational information, includes: acquiring laboratory calibration data of the satellite in its dawn-dusk orbit before flight; converting the observational information into radiance values ​​based on the laboratory calibration data; and performing a logarithmic scaling transformation on the radiance values ​​to obtain the radiance data.

[0033] As an optional implementation, to reduce the impact of image brightness non-uniformity on image contrast, step 5 includes the following steps: performing panoramic observation fitting on the corrected initial observation image to obtain a two-dimensional curved background function. Based on the two-dimensional curved background function, determining the ratio of each pixel to the background value in the corrected initial observation image to obtain a homogenization result. Extracting an enhanced display interval from the homogenization result. Performing a linear scaling transformation on the enhanced display interval to obtain the corrected observation image. The range of the enhanced display interval can be [0.8, 1.2].

[0034] As an optional implementation, to reduce the impact of changes in the solar zenith angle on the extraction of details such as clouds in the observed image, the process of constructing a fitting curve for the solar zenith angle based on radiance data in step 4 above may include: determining the radiance value of the solar zenith angle of each camera in the multi-camera low-light imager within a third predetermined angle range based on radiance data, and determining the average radiance value of each angle of the multi-camera low-light imager within the third predetermined angle range. A fitting curve for the solar zenith angle is then constructed based on the average radiance values ​​of each angle. Based on this, the process of correcting the initial observed image using the fitting curve for the solar zenith angle in step 4 to obtain a corrected initial observed image includes: using the fitting curve for the solar zenith angle to determine the relative difference between the initial observed image and the solar zenith angle at a second predetermined angle. The initial observed image is then corrected based on the relative difference to obtain the corrected initial observed image.

[0035] The process of constructing the fitting curve of the solar zenith angle based on radiance data in step 4 above can further include: determining the radiance value of the solar zenith angle of each camera in the multi-camera low-light imager within a fourth set angle range based on radiance data, and determining the average radiance value of each angle of the multi-camera low-light imager within the fourth set angle range. A fitting curve of the solar zenith angle is constructed based on the average radiance value of each angle. Based on this, the process of correcting the initial observation image based on the fitting curve of the solar zenith angle in step 4 to obtain the corrected initial observation image includes: using the fitting curve of the solar zenith angle to determine the relative difference of the initial observation image with respect to the solar zenith angle at a third set angle. The initial observation image is corrected based on the relative difference to obtain the corrected initial observation image.

[0036] By implementing steps 1 to 5 above, this application solves the problem of camera response differences affecting imaging consistency, and ultimately obtains uniform and consistent corrected observation images.

[0037] In an exemplary embodiment, for a low-light imaging observation image of a certain sea area on a certain morning acquired by a multi-camera low-light imager on a morning-evening orbit satellite, the consistency processing is performed using steps 1 to 5 provided in the above embodiment to obtain a uniform and consistent corrected observation image.

[0038] The satellite payload in the dawn-dusk orbit utilizes a space-based back-illuminated CCD array (i.e., a multi-camera low-light imager) with high detection sensitivity. Each camera is equipped with high-brightness and low-brightness imaging channels; the former is mainly used for daytime imaging, and the latter for nighttime imaging. The multiple settings for each channel can accommodate a wider dynamic range of observations. The observation information used in this embodiment is satellite L1 level data, mainly including low-light channel radiance values ​​(i.e., radiance values), imaging time, latitude and longitude, solar zenith angle, and other information.

[0039] Images of a certain sea area on a certain morning, obtained by the multi-camera low-light imager aboard a dawn-dusk orbit satellite, show the low-brightness and high-brightness channels respectively. Figure 2 As shown in parts (a) and (b), the radiance of the observed images in the low-brightness and high-brightness channels varies with the solar zenith angle as follows: Figure 3 As shown in parts (a) and (b), the radiance of the combined low-brightness and high-brightness channel observation image (i.e., the low-light imaging observation image) varies with the solar zenith angle as follows: Figure 3 As shown in section (c). For the observed image data, sunlight comes from the upper right of the image, with a solar zenith angle between 73 and 112 degrees; moonlight comes from the lower left of the image, with a lunar zenith angle between 44 and 82 degrees. The day was near a full moon in the lunar calendar, and information such as cloud cover could be obtained from the reflected moonlight at night. Figure 2 and Figure 3 It is evident that the observation location was near the twilight transition zone. Firstly, there was a significant non-uniformity between the low-brightness and high-brightness channels. Secondly, there were significant differences in the high-brightness channel between different cameras. Figure 2 In some areas, low-brightness and high-brightness channels are imaged simultaneously. However, the low-brightness channel tends to be too bright in response to sunlight, while the high-brightness channel is not sensitive to nighttime conditions. In business data processing systems, this often results in "black blocks" due to limited information content or "white blocks" due to overcorrection.

[0040] Depend on Figure 3 The observed radiance values ​​vary with the solar zenith angle, showing that the low-brightness channel at night primarily falls within the range of [-9, -4], indicating a dynamic range greater than four orders of magnitude. The high-brightness channel mainly falls within the range of [-6, -2], with a dynamic range of approximately four orders of magnitude, and the payload response exhibits significant nonlinearity at solar zenith angles above 95 degrees. This demonstrates that due to the limited dynamic range of a single-channel payload observation, dual-channel observation is necessary to achieve full-scene stitching imaging, but this also introduces significant inconsistencies at the stitching points. Clearly, when the dynamic range of the entire observation area exceeds the dynamic range of the detailed features of the local terrain, it becomes difficult to clearly display detailed changes in clouds, the land surface, and other features.

[0041] Depend on Figure 3It is evident that large dynamic range gain adjustment and nonlinear response processing in the region above the 92-degree solar zenith angle are prerequisites for extracting information such as clouds. Due to the unique design of the payload in this application, directly applying the processing methods of other payloads may result in significant deviations. For example, when the HNCC processing method is directly applied to the low-light image display in this embodiment, the daytime high-brightness channel is excessively bright near the 92-degree zenith angle. This is because within the above solar zenith angle range, the scene radiance is located at the boundary of the dynamic range of the high-brightness channel response, and its sensitivity decays, exhibiting nonlinear changes. Using the fixed gain processing of HNCC easily leads to overcorrection. Furthermore, due to the differences in the response characteristics of different cameras, there are significant differences at the stitching point between the third and fourth cameras in the high-brightness channel, affecting the subsequent processing and correction effects.

[0042] The images of the same area observed at the stitching point between the 3rd and 4th cameras are as follows: Figure 4 As shown in parts (a) and (b), the variation of radiance values ​​observed by the third and fourth cameras at the junction for the same area with the solar zenith angle is as follows. Figure 4 As shown in part (c) (x-axis is the angle of the solar zenith angle, y-axis is the logarithm of the radiance value, data represents, data1 represents, and fitted curve represents the fitted quadratic nonlinear curve), when the solar zenith angle is greater than 85 degrees, the radiance value of the 4th camera is significantly higher than that of the 3rd camera, and the difference increases with the increase of the solar zenith angle. This difference is mainly due to the nonlinear change in the sensitivity of the high-brightness channel of the 4th camera to lower radiation.

[0043] To reduce the impact of inconsistencies between cameras in a multi-camera low-light imager on the display of subsequent observed images, a correction method based on the solar zenith angle was established, namely step 2 in the above embodiment. Utilizing the overlapping observation area between adjacent cameras (taking the 3rd and 4th cameras as examples in this embodiment), the radiance value of the overlapping area between the 3rd and 4th cameras changes with the solar zenith angle as follows: Figure 4As shown in section (c). Using step 2 of the above embodiment, the inconsistency in the response between two adjacent cameras in the range where the solar zenith angle is greater than 85 degrees is addressed as follows: ① Calculate the observed mean solar zenith angle (i.e., mean radiance) for each of the two cameras based on the observation information. ② Fit the observed mean solar zenith angle of the first camera (range 85-87 degrees) and the observed mean solar zenith angle of the second camera (range 91-93 degrees) to obtain a quadratic nonlinear curve, which serves as the normal response reference curve. ③ Calculate the difference between the observed mean solar zenith angle of each camera and the normal response reference curve. ④ Perform a quadratic fitting of the difference value calculated in step ③ onto the solar zenith angle of each camera to obtain the correction functions for both cameras. ⑤ Based on the correction functions of each camera, perform correction processing on the low-light imaging observation image using a first set angle (85 degrees in this embodiment) as a reference to obtain the initial observation image. After correcting the low-light imaging observation image of a certain sea area on a certain morning using steps ① to ⑤ above, and comparing it with the observation results of the same camera at different settings, the relative difference in radiance after taking the logarithm decreased from 57% to 3.5%.

[0044] To improve the uniformity of low-light imaging observation images in the twilight transition zone, this application proposes an enhanced display method based on adaptive correction of the solar zenith angle and background normalization processing (i.e., steps 3-5 in the above embodiments). The processing flow for both low-brightness and high-brightness channels is as follows: Figure 5 As shown in the figure. This embodiment refers to the method as Adjusted High Near-Constant Contrast (AHNCC), which mainly consists of three parts: logarithmic transformation of radiance data, solar zenith angle correction, and homogenization processing. Solar zenith angle correction primarily aims to reduce the impact of variations in the solar zenith angle on the extraction of detail information such as clouds, while homogenization processing primarily aims to reduce the impact of image brightness non-uniformity on image contrast.

[0045] The logarithmic transformation process for radiance data includes data acquisition, radiance calculation, and logarithmic scaling: ① Data acquisition: This mainly involves acquiring observational information such as longitude, latitude, and solar zenith angle, as well as obtaining laboratory calibration data from the pre-flight phase-of-day orbit satellite. ② Radiance calculation: Based on the pre-flight laboratory calibration data, the observational information is converted into radiance values. ③ Logarithmic scaling: The radiance values ​​are taken as the logarithm to base 10, with a value range of [-11, -4], to obtain the radiance data.

[0046] The solar zenith angle correction process includes curve fitting, correction formula establishment, and linear scaling: ① Curve fitting: Based on the radiance data, determine the radiance value of the solar zenith angle within a third set angle range (90°~105° in this embodiment), and calculate the average radiance value for each solar zenith angle to construct a fitting curve for the solar zenith angle. ② Solar zenith angle difference calculation: Based on the fitting curve, determine the relative difference of the initial observation image with respect to the solar zenith angle at a second set angle (105° in this embodiment). ③ Use the relative difference of the solar zenith angle to perform nonlinear correction on the initial observation image to obtain the corrected initial observation image.

[0047] Image homogenization processing includes background surface fitting, local perturbation extraction, and standardization calculation: ① In background surface fitting, a panoramic observation fitting is performed on the corrected initial observation image to obtain a two-dimensional surface background function. ② The ratio of each pixel in the corrected initial observation image to the background (i.e., each pixel divided by the background value) is calculated based on the two-dimensional surface background function to obtain the homogenization processing result. ③ The enhanced display interval is truncated from the homogenization processing result (in this embodiment, values ​​between [0.8, 1.2] are truncated), and the augmented reality interval is linearly scaled using a linear function to map it to the grayscale value interval of [0, 255] for display, thus obtaining the corrected observation image.

[0048] The distribution of a low-light imaging observation image of a certain sea area on a certain morning, and the corrected observation image obtained after correction processing, as a function of the solar zenith angle is shown below. Figure 6 As shown ( Figure 6 Part a shows the distribution of the high-brightness channel radiance data after logarithmic transformation as a function of the solar zenith angle. Figure 6 Part b shows the distribution of the high-brightness channel after solar zenith angle correction as the solar zenith angle changes. Figure 6 Part c shows the distribution of the highlight channel after homogenization as the solar zenith angle changes. Figure 6 The d-part represents the distribution of the low-brightness channel radiance data after logarithmic transformation as a function of the solar zenith angle. Figure 6 Part e shows the distribution of the low-brightness channel after solar zenith angle correction as the solar zenith angle changes. Figure 6 The f-part represents the distribution of the low-brightness channel after homogenization as a function of the solar zenith angle. The frequency distribution of the observation data in this low-light imaging observation image and the corrected observation image obtained after correction processing is shown below. Figure 7 As shown ( Figure 7 Part a is a frequency graph of the radiance data of the high-brightness channel after logarithmic transformation. Figure 7 Part b is the frequency graph of the data after the highlight channel has been uniformized. Figure 7 Part c is a frequency graph of the low-brightness channel radiance data after logarithmic transformation. Figure 7Part d represents the frequency distribution of data after homogenization of the low-brightness channel. Before correction, the dynamic range of the displayed values ​​decreased after the logarithmic change of the radiance data, but it remained relatively bright when the solar zenith angle was small, and relatively dark when the solar zenith angle was large. That is, the dynamic range of the image is still significantly larger than the range of variation of cloud and ground object radiation, making it difficult to effectively extract relatively small details such as cloud cover from the image. After correcting the solar elevation angle for the low-brightness and high-brightness channels respectively, the images obtained for each channel are relatively uniform, and it is almost difficult to see the dependence on the solar zenith angle, but there are still obvious differences between the two channels. After homogenization processing for the high-brightness and low-brightness channels respectively, the results obtained for both channels show a normal distribution centered at 1, with most values ​​between 0.8 and 1.2. After linear scaling and mapping to the 0-255 grayscale value range, the image contrast is significantly enhanced, and the contrast of the two channels is comparable.

[0049] The impact of the correction process on image display quality, such as Figure 8 As shown ( Figure 8 Part a is the radiance diagram of the low-brightness channel. Figure 8 Part b is the observation image after correction of the solar zenith angle in the low-brightness channel. Figure 8 Part c is the observation image after the low-brightness channel has been homogenized. Figure 8 The d-part is the radiance diagram of the high-brightness channel. Figure 8 Part e is the observation image after correction of the solar zenith angle in the high-brightness channel. Figure 8 Part f is the observation image after the high-brightness channel has been homogenized. Obviously, after the solar zenith angle correction, the uniformity of the low-brightness and high-brightness channel observations is significantly improved; after the final standardization process, the image contrast and the consistency between the two channels are also significantly improved.

[0050] Comparison of stitched images processed by HNCC and processed by the consistency processing method proposed in this application (i.e., AHNCC) Figure 9 As shown ( Figure 9 Part a is the image after HNCC processing. Figure 9 Part b is the image after AHNCC processing. Figure 9 As can be seen, although the image obtained by HNCC has better contrast, there are obvious bright and dark areas in the image, and the contrast is not uniform across the entire scene. In contrast, the image processed by AHNCC shows relatively uniform contrast between the low-brightness and high-brightness channels. For the stitching area, the calculated average relative difference decreased from 41.32% in HNCC to 19.27% ​​in AHNCC. Figure 10 The image, after latitude and longitude projection correction, clearly shows the two typhoons and their structures. The transition between the low-brightness channel on the lower left and the high-brightness channel on the upper right after image stitching is relatively natural.

[0051] Based on the above comparison, the HNCC method generally adjusts parameters for the entire image, which can improve global contrast, but it cannot adaptively adjust parameters, making it difficult to achieve both regional and global optimization. Especially when applied to the multi-camera low-light imager on a satellite in a dawn-dusk orbit as described in this application, the low-end nonlinear response observed will result in overcorrection, making the image appear abnormally bright. The AHNCC method proposed in this application performs regional normalization processing to improve the global display effect with regional optimization, avoiding the "black blocks" caused by excessive darkness or the "white blocks" caused by excessive brightness.

[0052] To address the application requirements of operational weather forecasting for high-consistency, high-contrast low-light images from satellites in twilight orbits, this application proposes a consistency processing method for multi-camera low-light imaging in twilight orbits. This method addresses the impact of camera response differences on imaging consistency by correcting for inter-camera differences in the low-light imaging observation images. It also separates the low-brightness channel on the night side and the high-brightness channel on the day side near the terminator, resolving the inconsistency in the dynamic range of dual-channel observations. Furthermore, it addresses the non-uniformity of panoramic images through homogenization processing based on panoramic observation fitting. Finally, it determines radiance data based on observation information from twilight satellites, constructs a fitting curve based on the solar zenith angle, and sequentially performs radiance adjustment, solar zenith angle-based fitting curve correction, and homogenization processing on the initial observation image, ultimately obtaining a uniform and consistent corrected observation image.

[0053] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 11 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data related to a consistency processing method for multi-camera low-light imaging of satellites in a twilight orbit. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a consistency processing method for multi-camera low-light imaging of satellites in a twilight orbit.

[0054] Those skilled in the art will understand that Figure 11 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0055] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0056] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0057] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0058] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0059] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0060] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0061] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for consistency processing of low-light imaging by multiple cameras on a satellite in a dawn-dusk orbit, characterized in that, include: Acquire low-light imaging observation images and satellite observation information from dawn / dusk orbits; The low-light imaging observation images were acquired by a multi-camera low-light imager mounted on a dawn-dusk orbit satellite; The low-light imaging observation image is subjected to difference correction between cameras in a multi-camera low-light imager to obtain the initial observation image; The radiance data is determined based on the observation information; A fitting curve for the solar zenith angle is constructed based on the radiance data; the initial observation image is corrected based on the fitting curve for the solar zenith angle to obtain a corrected initial observation image; The corrected initial observation image is then homogenized to obtain the corrected observation image.

2. The consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit according to claim 1, characterized in that, The low-light imaging observation image is subjected to difference correction between cameras in a multi-camera low-light imager to obtain an initial observation image, including: Based on the observation information, the average solar zenith angle of every two adjacent cameras in the multi-camera low-light imager is determined. Based on the observed average values ​​of the solar zenith angle within the first set angle range of the first camera and the second set angle range of the second camera in every two adjacent cameras, a quadratic nonlinear curve is obtained. Determine the difference between the observed mean of the solar zenith angle of the first camera and the quadratic nonlinear curve; Based on the difference value, the solar zenith angle of the two cameras is fitted twice to obtain the correction functions of the two cameras respectively; The low-light imaging observation image is corrected based on the correction function of each camera and a first set angle as a reference to obtain the initial observation image.

3. The consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit according to claim 1, characterized in that, The process of determining radiance data based on the observation information includes: Obtain laboratory calibration data for satellites in dawn-dusk orbit before their flight; Based on the laboratory calibration data, the observation information is converted into radiance values; The radiance value is logarithmically scaled to obtain the radiance data.

4. The consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit according to claim 1, characterized in that, The corrected initial observation image is homogenized to obtain a corrected observation image, including: A panoramic observation fit is performed on the corrected initial observation image to obtain a two-dimensional curved surface background function; Based on the two-dimensional curved background function, the ratio of each pixel to the background value in the corrected initial observation image is determined to obtain the homogenization processing result; Extract the enhanced display area from the homogenization result; The enhanced display range is subjected to a linear scaling transformation to obtain the corrected observation image.

5. The consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit according to claim 1, characterized in that, The process of constructing a fitting curve for the solar zenith angle based on the radiance data includes: Based on the radiance data, determine the radiance value of each camera in the multi-camera low-light imager at the solar zenith angle within the third set angle range, and determine the average radiance value of the multi-camera low-light imager at each angle within the third set angle range. A fitting curve for the solar zenith angle is constructed based on the average radiance values ​​at various angles.

6. The consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit according to claim 5, characterized in that, The initial observation image is corrected based on the fitted curve of the solar zenith angle to obtain a corrected initial observation image, including: The relative difference of the initial observation image with respect to the solar zenith angle at the second set angle is determined by using the fitted curve of the solar zenith angle. The initial observation image is corrected based on the relative difference to obtain the corrected initial observation image.

7. The consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit according to claim 2, characterized in that, The first set angle range is 85 degrees to 87 degrees.

8. The consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit according to claim 2, characterized in that, The second set angle range is 91 degrees to 93 degrees.

9. The consistency processing method for multi-camera low-light imaging of satellites in dawn-dusk orbit according to claim 4, characterized in that, The range of the enhanced display interval is [0.8, 1.2].

10. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the consistency processing method for low-light imaging of multi-camera satellites in twilight orbit as described in any one of claims 1-9.

Images