A relative radiation correction method for a marine remote sensing ultra-wide scanning camera

Abstract

The application discloses a kind of relative radiation correction methods of marine remote sensing ultra-large width scanning camera, belong to optical remote sensing quantification technical field.The correction method first calculates image lookup table correction coefficient, then the 0-level image to be corrected is divided into multiple regional blocks according to scanning column, and the same number of overlapping image elements is reserved in adjacent regional blocks;Based on image lookup table correction coefficient, each regional block is corrected to obtain the partition image after rough correction;The residual noise of scanning strip in each rough correction partition image is modified, and the modified partition image is spliced according to the original position on the 0-level image to obtain the whole scene image.The application can solve the problem of inconsistent relative radiation response of visible light band of new generation of marine satellite water color and water temperature scanner, significantly improve the non-uniformity of marine remote sensing image, and ensure the accuracy of marine water color quantitative product.

Description

Technical Field

[0001] This invention relates to a relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing, belonging to the field of quantitative optical remote sensing technology. Background Technology

[0002] Monitoring ocean color elements and sea surface temperature fields is a scientific basis for the rational development and utilization of marine resources and an important reference for global climate change research. The new generation of ocean color and temperature scanners utilizes a telescope-wide rotating scanning imaging system to achieve daily revisits of global ocean color environment and sea surface temperature. Compared to previous ocean color and temperature scanners launched by my country, the new generation scanner covers 18 spectral bands including ultraviolet / visible / near-infrared / shortwave infrared / mid- and longwave infrared, with a 14-bit quantization bit, and uses an 80-element parallel scanning method to achieve a 3000-kilometer swath width. In the satellite's level 0 image, the flight direction consists of several scanning strips, and the forward and reverse scanning markers for each strip can be read from auxiliary data. The scanning direction has 3320 columns, such as... Figure 1 As shown. While the ultra-large scanning field of view obtains high temporal resolution data, the 80-element response difference, the difference between forward and reverse scanning, and the characteristics that vary with the field of view bring about a striping phenomenon in the image flight direction. This temporal and spatial radiometric inconsistency coupling poses a great challenge to subsequent radiometric calibration and application product inversion.

[0003] Relative radiometric correction can eliminate radiation striping caused by inconsistencies in sensor pixel response. Currently, relative radiometric correction of satellite ultraviolet / visible / near-infrared / shortwave infrared on-orbit images mainly relies on statistical processing methods, such as histogram matching and moment matching. Statistical methods obtain correction coefficients by analyzing and statistically analyzing the mapping relationship between the radiation distribution of the reference image and the image to be processed. However, the correction results are affected by various factors such as differences in internal image features, correction benchmarks, and the selection of mathematical correction models. The dynamic range of land cover types in ultra-wide-swath marine imagery varies greatly, especially in images over the ocean which are dominated by high-response clouds and low-response sea, making it impossible to obtain a balanced radiation response of the detector to the ground. Therefore, improved relative radiometric correction methods are needed for ultra-wide-swath, multi-element parallel-scan marine remote sensing images. Summary of the Invention

[0004] The technical problem solved by this invention is to overcome the shortcomings of the prior art and propose a relative radiometric correction method for ultra-wide scanning cameras for marine remote sensing, which significantly improves the non-uniformity of marine remote sensing images and ensures the accuracy of quantitative marine color products.

[0005] The technical solution of this invention is:

[0006] A method for relative radiometric correction of an ultra-wide-swath scanning camera for marine remote sensing, comprising:

[0007] S1: Accumulate at least two cycles of Level 0 images based on the satellite revisit cycle; divide each Level 0 image into several regions according to the scan column, retain multiple overlapping pixels in each adjacent region, and calculate the image lookup table correction coefficients.

[0008] S2: For the level 0 image to be corrected, divide it into the same number of region blocks as in step S1 according to the scan columns, and retain the same number of overlapping pixels in each adjacent region block; based on the correction coefficients of the image lookup table, correct each region block to obtain the coarsely corrected partitioned image.

[0009] S3: Correct residual noise in the scan strips of each coarsely corrected partition image;

[0010] S4: Stitch the corrected partitioned images according to their original positions on the level 0 image to obtain the whole scene image.

[0011] Furthermore, in step S1, the method for calculating the image lookup table correction coefficients is as follows:

[0012] For each region block of each level 0 image, 160 pixels from both forward and reverse scans are used as statistical samples to calculate a histogram of the DN value of each pixel.

[0013] Calculate the cumulative histogram S based on the histogram of each cell's DN value. j j = 1, 2, ..., 160;

[0014] The cumulative histogram S of the 40th pixel is obtained by scanning forward. 40 For calibration reference G;

[0015] Establish the cumulative histogram S for each of the other pixels. j The mapping relationship between the reference G and the calibration standard G is used, and the set of the smallest non-zero difference between the two is used as the correction coefficients in the lookup table.

[0016] Furthermore, using 160 pixels from both forward and reverse scans as a statistical sample, a histogram of the DN value for each pixel was generated. The following processing was performed during the statistical analysis:

[0017] Remove pixels with a DN value of 0 from the entire row, meaning that all data within a scan strip containing that row are not included in the statistical sample;

[0018] Remove pixels with a DN value of 0 that were filled in the right side of the region image when the data was received by the ground system.

[0019] Furthermore, in step S3, residual noise in the scan strips of each coarsely corrected partition image is corrected using the following method:

[0020] Calculate the number of overlapping pixels between different scan columns of the forward and reverse scan strips.repeat ;

[0021] For each scan column, the first overlapping cell and the pixel are scanned at each scan. repeat All pixels between pixels are taken as the homogenization region, and a homogenization model is established for two adjacent scans for homogenization processing; wherein the homogenization model is:

[0022] img n,k_out (i) = img n,k_corr_up (81-pixel repeat )×(1-delt(i))+img n,k_corr_down (i)

[0023] ×delt(i)

[0024] In the formula, delt(i) is the correction factor for the i-th pixel, and img n,k_out (i) represents the DN value of the i-th pixel in the partitioned image after homogenization; img n,k_corr_up The image shows the pixel DN values ​​of the previous scan strip before homogenization in two adjacent scans. n,k_corr_down The pixel DN value of the subsequent scan strip between two adjacent scans before homogenization; i ranges from 1 to pixel. repeat .

[0025] Further, in step S3, the number of overlapping pixels in different scan columns between the forward and reverse scan strips is calculated. repeat The calculation method is as follows:

[0026] pixel repeat (col) = round(0.0000153 × col) 2 -0.050218×col+50.37989)

[0027] In the formula, round() is the rounding function, and col is the scan column number, with a value of 1 to 3320.

[0028] Furthermore, in the uniform light model, the correction factor delt(i) for the i-th pixel is:

[0029] delt(i) = i / pixel repeat .

[0030] Further, in step S4, the corrected partitioned images are stitched together according to their original positions on the level 0 image. The stitching method is as follows:

[0031] S4.1: Based on the DN value of the first region block in the level 0 image, the DN values ​​of the other region blocks are corrected sequentially.

[0032] S4.2: Perform uniform light processing on the seams of the overall corrected partitioned image;

[0033] S4.3: Combine the partitioned images after the light equalization process into a whole scene image according to their original positions.

[0034] Further, in step S4.1, the DN values ​​of the pixels in other regions are sequentially corrected as a whole. The overall correction method is as follows:

[0035] Based on the DN value of the first region block in the level 0 image, adjacent preceding and following regions are grouped together and the following operations are performed sequentially according to their original positions:

[0036] Based on the overlapping pixel DN values ​​of the preceding region block and the adjacent following region block, the quadratic term correction model is substituted to calculate the quadratic term P2, the linear term P1, and the zero-order term coefficient P0; where the quadratic term correction model is:

[0037] img n,k_out_repreat_L (ki)=P2×img n,k_out_repreat_R (ki)×img n,k_out_repreat_R (ki)+P1×img n,k_out_repreat_R (ki)+P0

[0038] In the formula, img n,k_out_repreat_L and img n,k_out_repreat_R These are the DN values ​​of the ki-th overlapping pixel in the m-1 adjacent regions and the ki-th overlapping pixel in the m-th region, respectively. m ranges from 2 to N, where N is the number of regions in the 0-level image divided by the scan columns; ki represents the overlapping pixel number, ranging from 1 to M, where M is the number of overlapping pixels in each region.

[0039] Based on the obtained quadratic term P2, linear term P1, and zero-order term coefficient P0, the DN value of the m-th region block is sequentially corrected as a whole:

[0040] img′ n,k_out_m =P2×img n,k_out_m ×img n,k_out_m +P1×img n,k_out_m +P0

[0041] In the formula, img′ n,k_out_m and img n,k_out_L These are the DN values ​​of the image after overall correction for the m-th region block and the DN values ​​of the image before overall correction for the m-th region block, respectively.

[0042] Further, in step S4.2, the seams of the overall corrected partitioned image are subjected to uniform light processing, specifically as follows:

[0043] img″n,k_out_L (ii) = img' n,k_out_m_repeat_R (ii+5)×(1-delt(ii))+img′ n,k_out_m_repeat_L (ii)×delt(ii)

[0044] In the formula, ii represents the ii-th overlapping pixel, taking values ​​from 1 to M, where M is the number of overlapping pixels in each region block; img″ n,k_out_L This represents the DN value of the overlapping pixels after uniform lighting at the seam of the previous block, img′ n,k_out_m_repeat_R This indicates the DN value of the overlapping pixels before homogenization at the seam of the previous block; img′ n,k_out_m_repeat_L represents the DN value of the overlapping pixels before homogenization at the seam of the next block; delt(ii) is the correction factor for the ii-th overlapping pixel.

[0045] Furthermore, in step S1, each adjacent region block retains 5 overlapping pixels.

[0046] The advantages of this invention compared to the prior art are:

[0047] (1) The correction method proposed in this invention adopts the theory of histogram matching in image relative radiometric correction. It is simple and convenient to implement on orbit, easy to deploy in ground processing systems, and applicable to various payloads such as push-broom and swing-broom.

[0048] (2) The correction method of the present invention takes into account the difference in response of scanning angle and scanning line direction. By designing a partitioning strategy, a stable correction reference element is determined, and the radiation response difference between adjacent scans and different fields of view is optimized based on the inter-scan geometric matching model and the inter-block overlap area, which significantly improves the image non-uniformity. Attached Figure Description

[0049] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0050] Figure 1 A schematic diagram of a Level 0 image from a new generation of ocean satellite color and temperature scanner;

[0051] Figure 2 This is a flowchart of the relative radiometric correction method for an ultra-wide swath scanning camera for marine remote sensing, according to an embodiment of the present invention.

[0052] Figure 3 The images shown are comparisons of the on-orbit marine satellite image zoning correction effects according to an embodiment of the present invention. 3(a) is the image at level 0 before correction, and 3(b) is the image at level 0 after correction.

[0053] Figure 4 The above are comparison charts of the spectral curves of the on-orbit marine satellite image partition correction according to an embodiment of the present invention. 4(a) is the spectral curve of the 0th-order flight direction before correction, and 4(b) is the spectral curve of the 0th-order flight direction after correction. Detailed Implementation

[0054] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0055] This invention proposes a relative radiometric correction method for ultra-wide-swath scanning cameras in marine remote sensing, such as... Figure 2 As shown, it includes:

[0056] Step (1): Correction Dataset Construction

[0057] Based on the satellite revisit cycle, at least two cycles of Level 0 imagery are accumulated as a lookup table correction coefficient calculation dataset (img). n , where n is the total number of level 0 images in the dataset.

[0058] Step (II) Correction Area Division

[0059] The new generation ocean satellite's Level 0 data scan direction is 3320 columns. High-energy flares are present in the images. Based on statistical analysis of different energy levels, each image in step (I) is divided into 6 parts according to the scan columns. n,k (k = 1, 2, 3…, 6) Radiometric correction was performed in blocks. To reduce the radiometric differences when stitching together different blocks after correction, after multiple experiments, five overlapping pixels were retained in each adjacent region during block division, namely 1-605, 601-1105, 1101-1605, 1601-2205, 2201-2805, and 2801-3320. By adjusting the number of blocks and overlapping pixels, the computational load and correction effect were balanced.

[0060] Step (3): Regional Histogram Statistics

[0061] Histogram statistics of pixel DN values ​​were performed for each of the six blocks divided into images. First, the response differences between forward and reverse scans were considered, using 160 pixels from both scans as the statistical sample to calculate the histogram of DN values ​​for each pixel. Second, pixels with an entire row of DN values ​​of 0, possibly due to image transmission or decoding processes, were removed. That is, when a row of 0 pixels was found, all data within the scan strip (80 pixels) containing that row were excluded from the statistical sample. Finally, columns of 0 pixels filled in by the ground system when receiving data on the right side of the image were removed.

[0062] Step (IV): Constructing an image lookup table for correction coefficients

[0063] Calculate the cumulative histogram S for the histogram statistics of different blocks in step (iii). i (i = 1, 2, ..., 160). After analyzing massive amounts of on-orbit data to determine the variations between linear scanning directions, the cumulative histogram of the 40th pixel of the positive scan, which exhibits the most stable pixel radiometric characteristics, was used as the correction benchmark. Subsequently, S... j The mapping relationship between G and G (i.e., calculating the difference between the two) is used, and the group with the smallest non-zero difference is used as the correction coefficient in the lookup table. Through extensive analysis, the 40th element signal is found to be the most stable; therefore, the statistical histogram of the 40th element is a good correction benchmark for imaging performance.

[0064] Step (5): Regional Image Correction

[0065] Based on the principle of histogram matching, the six regions of the image are corrected according to the lookup table constructed in step (iv), resulting in the coarsely corrected partitioned image (img). n,k_corr .

[0066] Step (VI) Correction of Residual Noise in Scan Strips

[0067] To address the residual noise between different scan strips in the image, homogenization processing is performed on overlapping pixels between two scan strips, specifically as follows:

[0068] First, based on statistics of on-orbit images, a model for calculating the number of overlapping pixels in different scan columns between forward and reverse scan strips is constructed:

[0069] pixel repeat (col) = round(0.0000153 × col) 2 -0.050218×col+50.37989) (1)

[0070] In the formula, pixel repeat(col) represents the number of overlapping pixels in the scan column col of the forward and reverse scan strips. round() is the rounding function, and col is the scan column number, with values ​​ranging from 1 to 3320. The coefficients are stability coefficients obtained after analyzing a large number of on-orbit images.

[0071] Then, scan the first overlapping cell and the pixel... repeat All pixels between pixels are considered as homogenization regions, and homogenization models are established for two adjacent scans:

[0072] img n,k_out (i) = img n,k_corr_up (81-pixel repeat )×(1-delt(i))+img n,k_corr_down (i)

[0073] ×delt(i) (2)

[0075] In the formula, i = 1, 2, 3, ..., pixel repeat delt(i) is the correction factor for this pixel, img n,k_corr and img n,k_out These represent the DN values ​​of the pixels in the coarsely corrected and homogenized partitioned images, respectively. The subscripts up and down indicate the preceding and following scan stripes, respectively.

[0076] delt(i) = i / pixel repeat (3)

[0077] Step (VII) Correction Area Splicing Process

[0078] The six corrected image patches are stitched together. First, based on the five overlapping pixels of the previous region, a quadratic correction model for the overlapping pixels of the subsequent region is established, specifically as follows:

[0079] img n,k_out_repreat_L (ki)=P2×img n,k_out_repreat_R (ki)×img n,k_out_repreat_R (ki)+P1×img n,k_out_repreat_R (ki)+P0(4)

[0080] In the formula, P2, P1, and P0 are the coefficients of the quadratic, linear, and zero-order terms, respectively; ki represents the overlapping pixel number, with values ​​of 1, 2, 3, 4, and 5; img n,k_out_repreat_L and img n,k_out_repreat_R These are the overlapping pixel DN values ​​of two adjacent blocks after homogenization in step (six). The subscript L indicates that the pixel DN value belongs to the previous block, and the subscript R indicates that the pixel DN value belongs to the next block.

[0081] Then, based on the polynomial coefficients of the quadratic correction model, each subsequent region is corrected as a whole, using the previous region as a reference, to eliminate color differences caused by independent correction of each zone.

[0082] img′ n,k_out_m =P2×img n,k_out_m ×img n,k_out_m +P1×img n,k_out_m +P0 (5)

[0083] In the formula, img′ n,k_out_m and img n,k_out_L The values ​​are the image DN values ​​before and after different block splicing processes in step (VI), respectively, where m = 2, 3, 4, 5, 6.

[0084] Finally, uniform lighting processing is performed on the seams of the segmented images, and a correction model is established based on the values ​​of the last 5 elements of the previous region and the first 5 elements of the next region before correction:

[0085] img″ n,k_out_L (ii) = img' n,k_out_m_repeat_R (ii+5)×(1-delt(ii))+img′ n,k_out_m_repeat_L (ii)×delt(ii) (6)

[0086] In the formula, ii=end, end-1,..., end-4, img′ n,k_out_L 、img″ n,k_out _ L This represents the image after uniform lighting at the stitching seam of the previous section, img′ n,k_out_m_repeat_R This represents the DN value of the previous section seam before uniform light distribution, img′ n,k_out_m_repeat_L The value represents the DN value of the overlapping pixel before homogenization in the next partition. After multiple homogenization experiments, the delt(ii) values ​​of each overlapping pixel were determined to be 0.9, 0.7, 0.5, 0.3, and 0.1, respectively, where ii represents the position of the overlapping pixel and takes values ​​of 1, 2, 3, 4, and 5.

[0087] The six areas with corrected pixel values ​​at the stitching points were combined into a complete image based on their original positions.

[0088] img n_new =∪img″ n,k_out_L

[0089] In summary, the corrected whole-view image is obtained through the method proposed in this invention. Subsequent geometric processing by other professionals based on the whole-view image yields a spherical surface image that closely resembles the Earth's surface.

[0090] Figure 3The above is a comparison of the on-orbit marine satellite image zoning correction effect according to an embodiment of the present invention. 3(a) is the level 0 image before correction. The ocean in the center of the image field of view, which should be uniform and continuous, can be seen with obvious scanning stripes. After correction by the above method, the stripe effect of the level 0 image in 3(b) is significantly reduced.

[0091] Figure 4 The above is a comparison of the spectral curves of the on-orbit ocean satellite image partition correction according to an embodiment of the present invention. 4(a) is the curve of the change of DN value in the flight direction of the 0th level image before correction. The change of ocean DN value in the selected area shows a regular non-uniform change. 4(b) is the spectral curve of the ocean uniformly in the flight direction of the 0th level after correction. It can be seen that the ocean DN in different scanning bands has become smooth.

[0092] The embodiments described above are merely preferred embodiments of the present invention. Ordinary variations and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for relative radiometric correction of an ultra-wide-swath scanning camera for marine remote sensing, characterized in that, include: S1: Accumulate at least two cycles of Level 0 images based on the satellite revisit cycle; For each level 0 image, the area is divided into several regions according to the scanning column. Multiple overlapping pixels are retained in each adjacent region, and the image lookup table correction coefficient is calculated. S2: For the level 0 image to be corrected, divide it into the same number of region blocks as in step S1 according to the scan columns, and retain the same number of overlapping pixels in each adjacent region block; based on the correction coefficient of the image lookup table, correct each region block to obtain the coarsely corrected partitioned image. S3: Correct residual noise in the scan strips of each coarsely corrected partition image; S4: Stitch the corrected partitioned images according to their original positions on the level 0 image to obtain the whole scene image.

2. The relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 1, characterized in that, In step S1, the method for calculating the image lookup table correction coefficients is as follows: For each region block of each level 0 image, 160 pixels from both forward and reverse scans are used as statistical samples to calculate a histogram of the DN value of each pixel. Calculate the cumulative histogram S based on the histogram of each cell's DN value. j j = 1, 2, ..., 160; The cumulative histogram S of the 40th pixel is obtained by scanning forward. 40 For calibration reference G; Establish the cumulative histogram S for each of the other pixels. j The mapping relationship between the reference G and the calibration standard G is used, and the set of the smallest non-zero difference between the two is used as the correction coefficients in the lookup table.

3. The relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 2, characterized in that, Using 160 pixels from both forward and reverse scans as a statistical sample, a histogram of the DN value for each pixel was generated. The following processing was performed during the statistical analysis: Remove pixels with a DN value of 0 from the entire row, meaning that all data within a scan strip containing that row are not included in the statistical sample; Remove pixels with a DN value of 0 that were filled in the right side of the region image when the data was received by the ground system.

4. The relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 1, characterized in that, In step S3, residual noise in the scan strips of each coarsely corrected partition image is corrected using the following method: Calculate the number of overlapping pixels between different scan columns of the forward and reverse scan strips. repeat ; For each scan column, the first overlapping cell and the pixel are scanned at each scan. repeat All pixels between pixels are taken as the homogenization region, and a homogenization model is established for two adjacent scans for homogenization processing; wherein the homogenization model is: img n,k_out (i)=img n,k_corr_up (81-pixel repeat )×(1-delt(i))+img n,k_corr_down (i) ×delt(i) In the formula, delt(i) is the correction factor for the i-th pixel, and img n,k_out (i) represents the DN value of the i-th pixel in the partitioned image after homogenization; img n,k_corr_up The image shows the pixel DN values ​​of the previous scan strip before homogenization in two adjacent scans. n,k_corr_down The pixel DN value of the subsequent scan strip between two adjacent scans before homogenization; i ranges from 1 to pixel. repeat .

5. The relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 4, characterized in that, In step S3, the number of overlapping pixels in different scan columns between the forward and reverse scan strips is calculated. repeat The calculation method is as follows: pixel repeat (col)=round(0.0000153×col 2 -0.050218×col+50.37989) In the formula, round() is the rounding function, and col is the scan column number, with a value of 1 to 3320.

6. The relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 4, characterized in that, In the uniform light model, the correction factor delt(i) for the i-th pixel is: delt(i)=i / pixel repeat 。 7. The relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 1, characterized in that, In step S4, the corrected partitioned images are stitched together according to their original positions on the level 0 image. The stitching method is as follows: S4.1: Based on the DN value of the first region block in the level 0 image, the DN values ​​of the other region blocks are corrected sequentially. S4.2: Perform uniform light processing on the seams of the overall corrected partitioned image; S4.3: Combine the partitioned images after the light equalization process into a whole scene image according to their original positions.

8. The relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 7, characterized in that, In step S4.1, the DN values ​​of the pixels in other regions are corrected sequentially. The overall correction method is as follows: Based on the DN value of the first region block in the level 0 image, adjacent preceding and following regions are grouped together and the following operations are performed sequentially according to their original positions: Based on the overlapping pixel DN value of the previous region block and the overlapping pixel DN value of the adjacent next region block, the quadratic term correction model is substituted to calculate the quadratic term P2, the linear term P1 and the zero-order term coefficient P0. The quadratic term correction model is as follows: img n,k_out_repreat_L (ki)=P2×img n,k_out_repreat_R (ki)×img n,k_out_repreat_R (ki)+P1×img n,k_out_repreat_R (ki)+P0 In the formula, img n,k_out_repreat_L and img n,k_out_repreat_R These are the DN values ​​of the ki-th overlapping pixel in the m-1 adjacent regions and the ki-th overlapping pixel in the m-th region, respectively. m ranges from 2 to N, where N is the number of regions in the 0-level image divided by the scan columns; ki represents the overlapping pixel number, ranging from 1 to M, where M is the number of overlapping pixels in each region. Based on the obtained quadratic term P2, linear term P1, and zero-order term coefficient P0, the DN value of the m-th region block is sequentially corrected as a whole: img′ n,k_out_m =P2×img n,k_out_m ×img n,k_out_m +P1×img n,k_out_m +P0 In the formula, img′ n,k_out_m and img n,k_out_L These are the DN values ​​of the image after overall correction of the m-th region block and the DN values ​​of the image before overall correction of the m-th region block, respectively.

9. A relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 7, characterized in that, In step S4.2, the seams of the overall corrected partitioned image are subjected to uniform light processing, specifically as follows: img″ n,k_out_L (ii)=img′ n,k_out_m_repeat_R (ii+5)×(1-delt(ii))+img′ n,k_out_m_repeat_L (ii)×delt(ii) In the formula, ii represents the ii-th overlapping pixel, taking values ​​from 1 to M, where M is the number of overlapping pixels in each region block; img″ n,k_out_L This represents the DN value of the overlapping pixels after uniform lighting at the seam of the previous block, img′ n,k_out_m_repeat_R This indicates the DN value of the overlapping pixels before homogenization at the seam of the previous block; img′ n,k_out_m_repeat_L represents the DN value of the overlapping pixels before uniform lighting at the splicing seam of the next region block; delt(ii) is the correction factor for the ii-th overlapping pixel.

10. A relative radiometric correction method for an ultra-wide-swath scanning camera for marine remote sensing according to claim 1, characterized in that, In step S1, each adjacent region block retains 5 overlapping pixels.

Images