Radiation correction method and device for eliminating vignetting and striping of night light remote sensing image

Abstract

The embodiment of the application provides a radiation correction method and device for eliminating night light remote sensing image vignetting and stripes. The method comprises the following steps: first, acquiring an orbit image with an abnormally high brightness value, wherein the orbit image has n pixel columns; then, using a gray level co-occurrence matrix algorithm, extracting a plurality of image blocks with texture uniformity reaching a preset standard based on the orbit image; then, based on the plurality of image blocks, constructing n correction point sets corresponding to the n pixel columns, and using an i-th correction point set to train a nonlinear regression model as a correction model corresponding to the i-th pixel column; and then, using the n correction models corresponding to the n pixel columns to process an original night light remote sensing image to obtain a corrected night light remote sensing image. In this way, the vignetting and stripes in the faint light color and panchromatic image products can be eliminated.

Description

Technical Field

[0001] This application relates to the field of radiometric correction technology for remote sensing satellite images, and in particular to a radiometric correction method and apparatus for eliminating vignetting and stripes in nighttime light remote sensing images. Background Technology

[0002] The initial remote sensing satellite data can be traced back to Level 0 data, which is reconstructed, unprocessed instrument data and payload data with full resolution. Physical modeling is performed on this data using methods such as radiometric correction, geometric correction, and orthophoto projection to generate more advanced products.

[0003] The SDGSAT-1 satellite, developed by the Chinese Academy of Sciences' "Earth Big Data Science Engineering" pilot project, is the world's first scientific satellite dedicated to serving the United Nations 2030 Agenda for Sustainable Development. It carries three payloads: a low-light imager, a multi-spectral imager, and a thermal infrared imager. The low-light sensor uses pushbroom imaging to acquire 40m resolution color images and 10m resolution panchromatic images. A complete Class 0 image from the SDGSAT-1 satellite's high-resolution low-light sensor (Glimmer Imager for Urbanization, GIU) is captured by two cameras and a total of 16 charge-coupled devices (CCDs). Due to the non-uniform radiometric response of CCDs, brightness values ​​decay during imaging, leading to vignetting and stripe noise issues in CCD images.

[0004] With the continuous development of image processing technology, vignetting correction and stripe noise correction methods have been improved, providing technical support for radiometric correction of low-light Class 0 images. Currently, radiometric correction of Class 0 multispectral images largely utilizes sideslip data; however, GIU imaging makes it difficult to acquire sideslip data, and normal low-light images have low pixel brightness values, making it difficult to extract useful radiometric information. Based on existing methods, it is difficult to determine the size of the vignetting region at the CCD edge. Stripe noise exists in every Class 0 CCD image. Current denoising methods such as Fourier denoising, wavelet denoising, moment matching, variational methods, and low-rank matrix methods are suitable for low-resolution, small-scene remote sensing images, but are difficult to apply to the correction of SDGSAT-1 low-light orbital images. Vignetting and stripe phenomena still exist in existing SDGSAT-1 low-light color and panchromatic image products. Solving these problems is key to improving the radiometric correction quality of domestically produced high-resolution low-light images. Summary of the Invention

[0005] This application provides a radiometric correction method for eliminating vignetting and stripes in nighttime light remote sensing images, thereby eliminating vignetting and stripes in remote sensing satellite images and making the remote sensing satellite images clearer.

[0006] According to a first aspect, embodiments of this application provide a radiometric correction method for eliminating vignetting and stripes in nighttime light remote sensing images. The method includes: acquiring an orbital image with abnormally high brightness values, the orbital image having n pixel columns; employing a gray-level co-occurrence matrix algorithm; extracting several image blocks based on the orbital image whose texture uniformity meets a preset standard; constructing n correction point sets corresponding to the n pixel columns based on the several image blocks; training a nonlinear regression model using the i-th correction point set as a correction model corresponding to the i-th pixel column; and processing the original nighttime light remote sensing image using the n correction models corresponding to the n pixel columns to obtain a corrected nighttime light remote sensing image.

[0007] In one embodiment, acquiring an orbital image with abnormally high brightness values ​​includes: selecting the orbital image with abnormally high brightness values ​​as a calibration correction dataset.

[0008] In one embodiment, a gray-level co-occurrence matrix algorithm is used to extract several image blocks with texture uniformity reaching a preset standard based on the orbital image. This includes: taking the center pixel column of the orbital image as a reference, extracting multiple image blocks with a width of 100 pixels and a height of 50 pixels by taking 50 pixels to the left and right as the width; calculating the entropy and inverse moment of each image block in the extracted multiple image blocks based on the gray-level co-occurrence matrix algorithm; and classifying the image blocks in the multiple image blocks whose entropy is less than a first preset threshold and whose inverse moment is greater than a second preset threshold into the multiple image blocks.

[0009] In one embodiment, constructing n sets of correction points corresponding to n pixel columns based on the plurality of image blocks includes: for each image block, calculating the overall pixel mean of all pixels therein, and calculating the column pixel mean of any i-th column pixel in the n pixel columns; using the overall block pixel mean and the column pixel mean to form correction points and assign them to the i-th correction point set.

[0010] In one embodiment, the nonlinear model includes a gamma correction term.

[0011] In one embodiment, after processing the original nighttime light remote sensing image using n correction models corresponding to the n pixel columns to obtain the corrected nighttime light remote sensing image, the method further includes: adjusting pixels in the corrected nighttime light remote sensing image that are greater than a pixel brightness threshold to the pixel brightness threshold.

[0012] In one embodiment, after processing the original nighttime light remote sensing image using n correction models corresponding to the n pixel columns to obtain a corrected nighttime light remote sensing image, the method further includes: processing the corrected nighttime light remote sensing image using a moment matching algorithm to obtain a further corrected nighttime light remote sensing image.

[0013] According to a second aspect, a radiometric correction device for eliminating vignetting and stripes in nighttime light remote sensing images is provided, characterized in that the device comprises: an acquisition module for acquiring an orbital image with abnormally high brightness values, the orbital image having n pixel columns; an extraction module for extracting several image blocks with texture uniformity reaching a preset standard based on the orbital image using a gray-level co-occurrence matrix algorithm; a construction module for constructing n correction point sets corresponding to the n pixel columns based on the several image blocks; a training module for training a linear regression model using the i-th correction point set as a correction model corresponding to the i-th pixel column; and a first correction module for processing the original nighttime light remote sensing image using the n correction models corresponding to the n pixel columns to obtain a corrected nighttime light remote sensing image.

[0014] In one embodiment, the acquisition module is specifically used to: select the orbital image with abnormally high brightness values ​​as a calibration correction dataset.

[0015] In one embodiment, the extraction module is specifically used to: extract multiple image blocks with a width of 100 pixels and a height of 50 pixels by taking 50 pixels to the left and right as the width of the center pixel column of the orbital image as a reference; calculate the entropy and inverse moment of each image block in the multiple extracted image blocks based on the gray-level co-occurrence matrix algorithm; and classify the image blocks in the multiple image blocks whose entropy is less than a first preset threshold and whose inverse moment is greater than a second preset threshold into the plurality of image blocks.

[0016] In one embodiment, the construction module is specifically used to: calculate the overall pixel mean of all pixels in each image block, and calculate the column pixel mean of any i-th column of pixels in the n pixel columns; and use the overall block pixel mean and column pixel mean to form correction points and include them in the i-th correction point set.

[0017] In one embodiment, the apparatus further includes: a second correction module, used to process the corrected nighttime light remote sensing image using a moment matching algorithm to obtain a further corrected nighttime light remote sensing image. In summary, the radiometric correction method and apparatus for eliminating vignetting and stripes in nighttime light remote sensing images disclosed in the embodiments of this specification first selects an orbital image with abnormally high brightness values, extracts an image block 100 pixels wide and 50 pixels high from the center of the orbital image, processes the orbital image using a gray-level co-occurrence matrix algorithm to obtain a texture uniformity module, then calculates the average value of the texture uniformity module and the column mean of the orbital image to obtain a correction point set, then trains a nonlinear regression model using the correction point set, and finally uses the nonlinear regression model to correct the remote sensing nighttime light orbital image, thereby eliminating vignetting and stripes in the remote sensing satellite image and making the remote sensing satellite image clearer. Attached Figure Description

[0018] Figure 1This is a schematic flowchart of a radiometric correction method for eliminating vignetting and stripes in nighttime light remote sensing images provided in an embodiment of this application;

[0019] Figure 2 This shows an orbital image with abnormally high brightness values;

[0020] Figure 3 This shows an example of a segmented image block;

[0021] Figure 4 Showing a raw, color red-band satellite Class 0 image;

[0022] Figure 5 This demonstrates the use of the method of the present invention for... Figure 4 The corrected image is the original level 0 image that has been corrected.

[0023] Figure 6 Showing a raw, color green band satellite Class 0 image;

[0024] Figure 7 This demonstrates the use of the method of the present invention for... Figure 6 The corrected image is the original level 0 image that has been corrected.

[0025] Figure 8 Showing a raw, color blue-band satellite Class 0 image;

[0026] Figure 9 This demonstrates the use of the method of the present invention for... Figure 8 The corrected image is the original level 0 image that has been corrected.

[0027] Figure 10 Showing a raw panchromatic satellite Class 0 image;

[0028] Figure 11 This demonstrates the use of the method of the present invention for... Figure 10 The corrected image is the original level 0 image that has been corrected.

[0029] Figure 12 This shows another original panchromatic satellite Class 0 image;

[0030] Figure 13 This demonstrates the use of the method of the present invention for... Figure 12 The corrected image is the original level 0 image that has been corrected.

[0031] Figure 14 This is a schematic diagram of the structure of a radiometric correction device for eliminating vignetting and stripes in nighttime light remote sensing images, provided in an embodiment of this application. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be described below with reference to the accompanying drawings.

[0033] In the description of the embodiments of this application, the words "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a specific manner.

[0034] In the description of the embodiments in this application, the term "and / or" is merely a description of the association relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, B existing alone, and A and B existing simultaneously. Furthermore, unless otherwise stated, the term "multiple" means two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals.

[0035] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and their variations all mean "including but not limited to," unless otherwise specifically emphasized.

[0036] This application provides a radiometric correction method for eliminating vignetting and stripes in nighttime light remote sensing images. This method enables precise, robust, and universal correction of vignetting and stripes caused by radiation in nighttime light remote sensing images. For example, the nighttime light remote sensing image may be a Class 0 image acquired by the GIU sensor of the SDGSAT-1 satellite.

[0037] Figure 1 This is one embodiment provided in this application. For example... Figure 1 As shown, the method includes the following steps S10-S40.

[0038] In step S10, an orbital image with abnormally high brightness values ​​is acquired, the orbital image having n pixel columns.

[0039] It is understandable that the abnormally high brightness value refers to a brightness value that is close to but less than the maximum brightness value that the CCD sensor can receive.

[0040] In some embodiments, orbits with abnormally high brightness values ​​are selected from the raw Level 0 data files of the SDGSAT-1 satellite on different dates, and the Level 0 images in the files are read as the correction dataset. It should be understood that the SDGSAT-1 satellite passes overhead around 9 PM local time, and before nightfall, the scattering of sunlight by the upper atmosphere causes the low-light images to exhibit uniform abnormally high brightness values. Figure 2 This invention presents an orbital image with abnormally high brightness values. Typically, such images are discarded because ground illumination information cannot be discerned. This invention cleverly uses them as uniform calibration field data for low-light images, solving the problem of obtaining effective sideslip data from low-light images. From this, a correction dataset can be obtained, which includes one or more orbital images with abnormally high brightness values.

[0041] In step S20, a gray-level co-occurrence matrix algorithm is used to extract several image blocks with uniform texture that meet a preset standard based on the orbital image. The sub-steps of this step are described below:

[0042] (1) Select the orbital images in the calibration dataset and extract multiple image blocks from the orbital images.

[0043] In some embodiments, the image located in the center of the orbital image is divided into image blocks of size 100 pixels wide and 50 pixels high. That is, the image width is n, and based on the n / 2-50th to n / 2+50th pixel columns, image blocks of size 100×50 are divided from the first row to the last row. Figure 3 This example shows a segmented image block, where the white-framed areas represent the segmented image blocks.

[0044] In one example, the orbital image is 500 pixels wide and 1000 pixels high. Based on columns 200 to 300, the image area from row 0 to row 49 is designated as the first image block; the image area from row 50 to row 99 is designated as the second image block, and so on, until the last complete image block is defined.

[0045] In other embodiments, multiple image blocks of predetermined size are randomly extracted from the orbital imagery.

[0046] (2) Based on the gray-level co-occurrence matrix algorithm, calculate the entropy and inverse difference moment of each of the multiple image blocks.

[0047] Specifically, entropy and inverse different moment (IDM) from the Gray-Level Co-occurrence Matrix (GLCM) algorithm are used as evaluation metrics for texture uniformity. For an introduction to GLCM, please refer to: Liu, Yongkun, et al. "A General Relative Radiometric Correction Method for Vignetting and Chromatic Aberration of Multiple CCDs: Take the Chinese Series of Gaofen Satellite Level-0 Images for Example." IEEE Transactions on Geoscience and Remote Sensing 60 (2022): 1-25.

[0048] The entropy and inverse difference of the image block can be calculated using formulas (1) and (2), respectively:

[0049]

[0050]

[0051] In equations (1) and (2), L represents the maximum number of gray levels in the image, and g(i,j) is the GLCM value of pixel (i,j).

[0052] In this way, the entropy and inverse difference value of each extracted image block can be calculated.

[0053] (3) The image blocks whose entropy is less than the first preset threshold and whose inverse moment is greater than the second preset threshold are classified into a number of image blocks whose texture uniformity meets the preset standard.

[0054] It should be understood that uniformly textured images have small entropy values ​​and large IDM values. Therefore, image blocks that simultaneously satisfy an entropy less than a first preset threshold (e.g., 0.03) and an inverse moment greater than a second preset threshold (e.g., 0.97) can be considered uniformly textured image blocks. In one example, the entropy and inverse moment of the image block are calculated, and image blocks with an entropy value less than 0.03 and an inverse moment greater than 0.97 are considered uniformly textured image blocks.

[0055] Based on the above, several image blocks with uniform texture can be extracted from the orbital image.

[0056] In step S30, based on the above-mentioned image blocks, a set of n correction points corresponding to n pixel columns is constructed.

[0057] Specifically, the orbital image contains n columns of pixels, where n is a natural number. Based on this, for the plurality of image blocks, firstly, the overall pixel mean of all pixels within the blocks is calculated, and then the column pixel mean of any i-th column of the n pixel columns is calculated, where i ∈ [1, n]. Then, using the overall block pixel mean and the column pixel mean as coordinate values, correction points are formed and assigned to the i-th correction point set. In this way, n correction point sets corresponding to the n pixel columns can be obtained.

[0058] It should be noted that the above mainly uses the processing of a certain orbital image as an example to introduce the acquisition of n correction point sets. In fact, the above processing can also be performed on other orbital images in the above correction dataset, so that the obtained correction points can be classified into n correction point sets.

[0059] From the above, we can obtain a set of n correction points.

[0060] In step S40, a nonlinear regression model is trained using the i-th correction point set as the correction model corresponding to the i-th pixel column.

[0061] By observing and analyzing the n correction point sets, it was found that the correction points have a linear relationship at low brightness values ​​and a nonlinear relationship at high brightness values. Therefore, it is proposed that, in one embodiment, a nonlinear function with added Gamma correction term is constructed, where the Gamma correction term can be found in reference [H. Farid, "Blind inverse gamma correction," IEEE transactions on image processing, vol. 10, no. 10, pp. 1428-1433, 2001]. Specifically, the vignetting and stripe noise correction parameters of each column (i-th column) are fitted using the Levenberg-Marquardt method (combining gradient method and Newton's method), which can be seen in the following formula (3):

[0062]

[0063] In equation (3), mean nov The mean value is the overall average pixel brightness of a uniform texture block. i ω, b, c, γ are the average pixel brightness values ​​of the image, n is the maximum number of gray levels in the image, and ω, b, c, γ are the required correction parameters.

[0064] In one example, the nonlinear regression model shown in formula (3) is trained using the i-th correction point set, and the trained nonlinear regression model is used as the correction model for the i-th column of pixels, thereby obtaining n correction models corresponding to n pixel columns.

[0065] In another embodiment, other types of nonlinear regression models besides those in formula (3) can be constructed and trained using n calibration point sets respectively to obtain n calibration models.

[0066] From the above, we can obtain n calibration models.

[0067] In step S50, the original nighttime light remote sensing image is processed using n correction models corresponding to the n pixel columns to obtain the corrected nighttime light remote sensing image. For example, the original nighttime light remote sensing image is a level 0 image.

[0068] In this step, each column of pixels in the corresponding position of the nighttime light remote sensing image is corrected based on n correction models. Specifically, the i-th correction model is used to correct the i-th column of pixels in the nighttime light remote sensing image. In this way, vignetting and most of the stripe noise in the original nighttime light remote sensing image can be corrected.

[0069] In one embodiment, based on the nighttime light remote sensing image corrected by a nonlinear regression model, pixels with a brightness value greater than the maximum brightness value that the CCD sensor can receive are adjusted to a pixel threshold to prevent overfitting.

[0070] In one embodiment, the corrected nighttime light remote sensing image is processed using a moment matching algorithm. In one example, the standard deviation, mean, standard deviation, and mean of each column of the corrected nighttime light remote sensing image are calculated, and residual stripe noise is eliminated using the moment matching method.

[0071]

[0072] In equation (4), σ is the corrected brightness value of pixel (i, j). j and m j are the standard deviation and mean of the j-th column, respectively, and m is the overall mean of the image.

[0073] The following are some application examples of correcting nighttime light remote sensing images using the method of this invention.

[0074] Table 1. Specific Applications of SDGSAT-1 Low-Light Grade 0 Image Radiometric Correction

[0075]

[0076] Table 1 shows the application example using the SDGSAT-1 satellite GIU sensor. Eight orbital images were selected, and calibration and validation set data were created. The calibration parameters obtained from the calibration set were directly applied to the validation set images, yielding good radiometric correction results.

[0077] Application Example 1: Select a color red band satellite image from May 21, 2022. The method of this invention can solve the problems of vignetting and stripe noise that appear in the image. See [link / reference]. Figure 4 and Figure 5 , Figure 4 Original Level 0 image, Figure 5 For images corrected using the method of this invention, the method of this application can eliminate vignetting and stripe noise problems in the images.

[0078] Application Example 2: A color green band satellite image from June 26, 2022, was selected. The method of this invention can solve the problems of vignetting and stripe noise that appear in the image. See [link / reference]. Figure 6 and Figure 7 , Figure 6 Original Level 0 image, Figure 7 For images corrected using the method of this invention, the method of this application can eliminate vignetting and stripe noise problems in the images.

[0079] Application Example 3: A color blue-band satellite image from July 2022 was selected. The method of this invention can solve the problems of vignetting and stripe noise that appear in the image. See [link / reference]. Figure 8 and Figure 9 , Figure 8 Original Level 0 image, Figure 9 For images corrected using the method of this invention, the method of this application can eliminate vignetting and stripe noise problems in the images.

[0080] Application Example 4: A panchromatic satellite image from May 21, 2022, is selected. The method of this invention can solve the problems of vignetting and stripe noise that appear. See [link / reference]. Figure 10 and Figure 11 , Figure 10 Original Level 0 image, Figure 11 For images corrected using the method of this invention, the method of this application can eliminate vignetting and stripe noise problems in the images.

[0081] Application Example 5: A panchromatic satellite image from July 8, 2022, is selected. The method of this invention can solve the problems of vignetting and stripe noise. See [link / reference]. Figure 12 and Figure 13 , Figure 12 Original Level 0 image, Figure 13 For images corrected using the method of this invention, the method of this application can eliminate vignetting and stripe noise problems in the images.

[0082] Corresponding to the above method, this specification also discloses a calibration device. Figure 14 This is a schematic diagram of the structure of a radiometric correction device for eliminating vignetting and stripes in nighttime light remote sensing images, as provided in an embodiment of this application. Figure 14 As shown, the device 1400 includes the following modules: an acquisition module 1410, used to acquire an orbital image with abnormally high brightness values, the orbital image having n pixel columns; an extraction module 1420, used to extract several image blocks with texture uniformity reaching a preset standard based on the orbital image using a gray-level co-occurrence matrix algorithm; a construction module 1430, used to construct n correction point sets corresponding to the n pixel columns based on the several image blocks; a training module 1440, used to train a linear regression model using the i-th correction point set as a correction model corresponding to the i-th pixel column; and a first correction module 1450, used to process the original night light remote sensing image using the n correction models corresponding to the n pixel columns to obtain a corrected night light remote sensing image.

[0083] In one embodiment, the extraction module 1420 is specifically used to: extract multiple image blocks from the orbital image; calculate the entropy and inverse moment of each image block in the multiple image blocks based on the gray-level co-occurrence matrix algorithm; and classify the image blocks in the multiple image blocks whose entropy is less than a first preset threshold and whose inverse moment is greater than a second preset threshold into the plurality of image blocks.

[0084] In one embodiment, the construction module 1430 is specifically used to: calculate the overall pixel mean of all pixels in each image block, and calculate the column pixel mean of any i-th column pixel in the n pixel columns; and use the overall pixel mean and column pixel mean to form correction points and include them in the i-th correction point set.

[0085] In one embodiment, the device 1400 further includes a second correction module 1460, used to process the corrected nighttime light remote sensing image using a moment matching algorithm to obtain a further corrected nighttime light remote sensing image.

[0086] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means, such as coaxial cable, optical fiber, digital subscriber line (DSL), or wireless means, such as infrared, wireless, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage and computing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape, etc.), an optical medium (such as a DVD, etc.), or a semiconductor medium, such as a solid-state disk (SSD), etc.

[0087] It is understood that the various numerical designations used in the embodiments of this application are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application.

Claims

1. A radiometric correction method for eliminating vignetting and stripes in nighttime light remote sensing images, characterized in that, The method includes: Acquire an orbital image with abnormally high brightness values, the orbital image having n pixel columns; the acquisition of the orbital image with abnormally high brightness values ​​includes: selecting the orbital image with abnormally high brightness values ​​as a calibration correction dataset; The gray-level co-occurrence matrix algorithm is used to extract several image blocks whose texture uniformity meets a preset standard based on the orbital image. Based on the aforementioned image blocks, n correction point sets corresponding to the n pixel columns are constructed; a nonlinear regression model is trained using the i-th correction point set as the correction model corresponding to the i-th pixel column; the nonlinear model includes a gamma correction term; The original nighttime light remote sensing image is processed using n correction models corresponding to the n pixel columns to obtain the corrected nighttime light remote sensing image.

2. The method according to claim 1, characterized in that, Using a gray-level co-occurrence matrix algorithm, several image blocks with texture uniformity reaching a preset standard are extracted from the orbital image, including: Based on the center pixel column of the orbital image, multiple image blocks with a width of 100 pixels and a height of 50 pixels are extracted by taking 50 pixels to the left and right as the width. Based on the gray-level co-occurrence matrix algorithm, the entropy and inverse difference moment of each image block in the extracted multiple image blocks are calculated; Image blocks whose entropy is less than a first preset threshold and whose inverse moment is greater than a second preset threshold are classified into the plurality of image blocks.

3. The method according to claim 1, characterized in that, The step of constructing a set of n correction points corresponding to the n pixel columns based on the plurality of image blocks includes: For each image block, calculate the overall pixel mean of all pixels therein, and calculate the column pixel mean of any i-th column pixel in the n pixel columns; Correction points are formed using the overall block pixel average and column pixel average, and are then assigned to the i-th correction point set.

4. The method according to claim 1, characterized in that, After processing the original nighttime light remote sensing image using n correction models corresponding to the n pixel columns to obtain the corrected nighttime light remote sensing image, the method further includes: In the corrected nighttime light remote sensing image, pixels with brightness values ​​greater than the pixel brightness threshold are adjusted to the pixel brightness threshold.

5. A radiometric correction device for eliminating vignetting and stripes in nighttime light remote sensing images, characterized in that, The device includes: An acquisition module is used to acquire orbital images with abnormally high brightness values, wherein the orbital images have n pixel columns; the acquisition of orbital images with abnormally high brightness values ​​includes: selecting the orbital images with abnormally high brightness values ​​as a calibration correction dataset. The extraction module is used to extract several image blocks with texture uniformity reaching a preset standard based on the orbital image using a gray-level co-occurrence matrix algorithm. The construction module constructs a set of n correction points corresponding to the n pixel columns based on the plurality of image blocks; The training module uses the i-th correction point set to train a linear regression model, which serves as the correction model corresponding to the i-th pixel column; the nonlinear model includes a gamma correction term. The first correction module processes the original nighttime light remote sensing image using n correction models corresponding to the n pixel columns to obtain the corrected nighttime light remote sensing image.

6. The apparatus according to claim 5, characterized in that, The extraction module is specifically used for: Based on the center pixel column of the orbital image, multiple image blocks with a width of 100 pixels and a height of 50 pixels are extracted by taking 50 pixels to the left and right as the width. Based on the gray-level co-occurrence matrix algorithm, the entropy and inverse difference moment of each image block in the extracted multiple image blocks are calculated; Image blocks whose entropy is less than a first preset threshold and whose inverse moment is greater than a second preset threshold are classified into the plurality of image blocks.

7. The apparatus according to claim 5, characterized in that, The building module is specifically used for: For each image block, calculate the overall pixel mean of all pixels therein, and calculate the column pixel mean of any i-th column pixel in the n pixel columns; Correction points are formed using the overall block pixel average and column pixel average, and are then assigned to the i-th correction point set.

8. The apparatus according to claim 5, characterized in that, The device further includes: The second correction module is used to process the corrected nighttime light remote sensing image using a moment matching algorithm to obtain a further corrected nighttime light remote sensing image.

Images