Fast relative radiometric calibration method for local strip noise of optical remote sensing satellite image

Abstract

The optical remote sensing satellite image local strip noise fast relative radiation calibration method belongs to the technical field of optical remote sensing images, solves the technical problem that the newly added vertical strip noise of the image cannot be quickly and effectively corrected due to the sudden change of the imaging state of the sensor of the optical remote sensing satellite during the on-orbit operation of the optical remote sensing satellite, and comprises the following steps: searching for a calibration image with a uniform ground object covering strip position; using original calibration coefficients to perform relative radiation correction on the image; calculating strip noise calibration coefficients based on the uniform area image; correcting for two cases of laboratory relative radiation calibration coefficients and on-orbit relative radiation calibration coefficients, and the updated calibration coefficients can adapt to the changed sensor imaging state, and effectively eliminate the newly added local strip noise of the image.

Description

Technical Field

[0001] This invention belongs to the field of optical remote sensing image technology. Background Technology

[0002] Relative radiometric correction is the process of eliminating or correcting various noises added to the radiation energy output by the sensor during optical remote sensing satellite imaging. It reduces or eliminates the response differences of various detection elements in the sensor and ensures that the response of the detection elements to the input radiance is uniform and consistent.

[0003] For optical remote sensing satellites using pushbroom imaging mode, existing relative radiometric correction methods are broadly classified into two categories based on algorithm principles: calibration methods and data model-based methods. Calibration methods include laboratory calibration, on-board calibration, site calibration, cross-calibration, and agile satellite yaw calibration. Laboratory calibration uses an integrating sphere as a standard light source, collecting imaging data from the camera under different input radiance levels to obtain laboratory relative radiometric calibration coefficients. However, as the satellite operates in orbit, the sensor's on-orbit status changes, and the effectiveness of the calibration coefficients gradually decreases. On-board calibration obtains calibration data by responding to the input energy of a built-in standard lamp or the sun, also offering high accuracy. However, it requires the satellite to carry corresponding calibration equipment, resulting in higher costs, and the calibration devices also experience aging issues. Site calibration, as an alternative calibration method, can use large areas of flat terrain with uniform reflectivity as the shooting reference, but it has stringent requirements regarding the shooting location and weather conditions, making it unsuitable for wide-format cameras. Cross-calibration uses high-precision satellite imagery data as a reference, which is low-cost, but it imposes constraints on the observation geometry and weather conditions at the imaging time of both the reference and target satellites, and requires similar spectral response ranges for matching. Yaw calibration rotates the camera by 90° so that each sensor can image the same ground features, thus achieving relative radiometric calibration of the sensor, but it places high demands on the satellite's maneuverability and imaging attitude stability.

[0004] Data model-based methods include frequency domain methods and spatial domain statistical methods. Frequency domain methods transform remote sensing images to the frequency domain and use different filters to eliminate noise, thereby correcting the consistency of image response. These methods include Mask homogenization, Retinex variational method, homomorphic filtering, and wavelet transform. However, these methods depend on the design and parameter selection of the filters and may not always obtain optimal image results, resulting in poor versatility.

[0005] Spatial domain statistical methods, including histogram matching and moment matching, are currently widely used and considered reliable. These methods establish correlations between pixel grayscale statistical information based on statistical analysis of large amounts of sample data. Most optical remote sensing satellites initially use laboratory relative radiometric calibration coefficients to perform relative radiometric correction on images. However, due to changes in the sensor's on-orbit condition, laboratory relative radiometric calibration coefficients cannot completely and effectively eliminate response differences between individual elements in the image. Therefore, during the satellite's on-orbit testing phase, histogram lookup tables for each element, i.e., on-orbit relative radiometric calibration coefficients, are typically calculated using histogram matching algorithms based on months of accumulated imaging data.

[0006] However, during satellite operation, sensor states often undergo abrupt changes, frequently manifesting as vertical stripe noise in localized areas of the image. If histogram matching algorithms are still used for relative radiometric correction in such cases, it requires re-accumulating imaging data, which is time-consuming and cannot quickly resolve the local failure of relative radiometric calibration coefficients. Consequently, the newly added local stripe noise in the image cannot be effectively eliminated in the short term. Summary of the Invention

[0007] The present invention aims to solve the technical problem that the newly added stripe noise in the image caused by sudden changes in sensor status during the on-orbit operation of optical remote sensing satellites cannot be quickly and effectively corrected.

[0008] A rapid relative radiometric calibration method for local stripe noise in optical remote sensing satellite imagery includes the following steps:

[0009] Step 1: For the problem area with stripe noise in the image, find uniformly imaged ground features with uniform reflectivity, so that the uniformly imaged ground feature covers the location of the stripe noise by more than 5 pixels on the left and right, and has a height of more than 1000 rows.

[0010] Step 2: Perform relative radiometric correction on the image using laboratory or in-orbit relative radiometric calibration coefficients to obtain the corrected image;

[0011] Step 3: Based on the uniform region selected in Step 1, correct the vertical stripe noise in the corrected image obtained in Step 2, and calculate the scaling coefficient coeff. i The formula is as follows:

[0012]

[0013] In the formula, The average gray level of the image over a uniform region. The average grayscale value of the i-th column of the uniform region image, where i is the sensor pixel column number. startcol i represents the leftmost column number of the uniform region. endcol The rightmost column number of the uniform region;

[0014] Step 4: If the sensor only has a laboratory relative radiometric calibration coefficient, then correct that coefficient:

[0015] The gain term in the laboratory relative radiation calibration coefficient is extracted and corrected, while the bias term remains unchanged. The formula is as follows:

[0016]

[0017] In the formula, Let be the original laboratory relative radiometric calibration coefficient for sensor pixel i. The laboratory relative radiometric calibration coefficient for sensor pixel i after correction;

[0018] If the sensor already has on-orbit relative radiometric calibration coefficients, then those coefficients should be corrected:

[0019] The mapping values ​​of each pixel in the on-orbit relative radiometric calibration coefficients at different gray levels are extracted and corrected accordingly, as shown in the following formula:

[0020]

[0021] In the formula, Let be the mapped gray value corresponding to the original on-orbit relative radiometric calibration coefficient of sensor pixel i when the gray value is j. Let $n$ be the mapped gray value corresponding to the on-orbit relative radiometric calibration coefficient of sensor pixel $i$ at gray value $j$, and $n$ be the sensor quantization bits. n -1 represents the sensor's saturation grayscale value.

[0022] Technical Effects: This invention can quickly update the relative radiometric calibration coefficients when the sensor state changes abruptly during the on-orbit operation of an optical remote sensing satellite, effectively eliminating newly added local stripe noise in the image. Based on the original calibration coefficients of the sensor, the correction value of the calibration coefficients is calculated by finding a small area of ​​uniform ground object imagery covering the location of stripe noise. The updated calibration coefficients can adapt to the changed sensor imaging state, with low time cost and easy implementation. Attached Figure Description

[0023] Figure 1 The image shown is an embodiment of the present invention after correction by the original relative radiometric calibration coefficient.

[0024] Figure 2 for Figure 1 A schematic diagram of the vertical strip noise effect after local stretching and enhancement of the white frame area of ​​the image.

[0025] Figure 3 This is the calibration coefficient curve for image stripe noise in an embodiment of the present invention.

[0026] Figure 4This is a comparison of the image correction results before (left) and after (right) calibration coefficient correction under the imaging condition of 48-level integral stages and 2x gain in an embodiment of the present invention.

[0027] Figure 5 This is a comparison of the image correction results before (left) and after (right) calibration coefficient correction under the imaging conditions of 24-level integration stages and 3 times gain in an embodiment of the present invention. Detailed Implementation

[0028] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] This embodiment provides a method for rapid relative radiometric calibration of local stripe noise in optical remote sensing satellite images, including the following steps:

[0030] Step 1: For the problem area with stripe noise in the image, find uniformly imaged ground features with uniform reflectivity, so that the uniformly imaged ground feature covers the location of the stripe noise by more than 5 pixels on the left and right, and has a height of more than 1000 rows.

[0031] Taking a certain type of remote sensing satellite as an example, due to the movement of the position of the extraneous objects attached to the sensor, new vertical strip noise appears in its panchromatic channel image. The satellite data management platform is used to select images with uniformly distributed areas of ground objects covering the strip positions. The imaging conditions are 48-level integral series and 2x gain.

[0032] Preferably, the uniformly imaged ground features are deserts or bodies of water.

[0033] Step 2: Perform relative radiometric correction on the image using laboratory or in-orbit relative radiometric calibration coefficients to obtain the corrected image.

[0034] like Figure 1 As shown, this is an image obtained after correction using the original on-orbit relative radiometric calibration coefficients, along with its local details. To make the image more visually appealing, as shown... Figure 2 As shown, for Figure 1 The local image within the white box has been stretched and enhanced, which is visible. Figure 2 There is a noticeable vertical stripe of noise in the left side of the image. Figure 2 The area highlighted in red is the uniform region selected for correcting strip noise.

[0035] Step 3: Based on the uniform region selected in Step 1, correct the vertical stripe noise in the corrected image obtained in Step 2, and calculate the scaling coefficient coeff. i The formula is as follows:

[0036]

[0037] In the formula, The average gray level of the image over a uniform region. Let i be the average gray level of the i-th column of the uniform region image, where i is the column number of the detector pixel. startcol i represents the leftmost column number of the uniform region. endcol The rightmost column number of the uniform region; such as Figure 3 As shown, this is the calibration coefficient curve calculated based on different cell column numbers.

[0038] Step 4: In this embodiment, since the sensor uses on-orbit relative radiometric calibration coefficients in the form of a histogram lookup table, the following steps are performed:

[0039] The mapping values ​​of each pixel in the on-orbit relative radiometric calibration coefficients at different gray levels are extracted and corrected accordingly, as shown in the following formula:

[0040]

[0041] In the formula, Let be the mapped gray value corresponding to the original on-orbit relative radiometric calibration coefficient of sensor pixel i when the gray value is j. Let $n$ be the mapped gray value corresponding to the on-orbit relative radiometric calibration coefficient of sensor pixel $i$ at gray value $j$, and $n$ be the sensor quantization bits. n -1 represents the sensor's saturation grayscale value.

[0042] like Figure 4 The image shown is a comparison of the image correction results before (left) and after (right) calibration coefficient correction for an image under the condition of 48-level integration series and 2x gain.

[0043] Furthermore, if the sensor has multiple imaging conditions, then change the imaging conditions and repeat steps one to four until the relative radiometric calibration coefficients are corrected and updated under all conditions.

[0044] Furthermore, the imaging conditions are integral series and gain.

[0045] In this embodiment, the stripe noise in the image is caused by the changing position of extraneous objects attached to the detector. Therefore, the changes in the detector's imaging state are not consistent at different integration levels, but they are not related to the imaging gain. Therefore, the histogram lookup table corresponding to different gains at the same integration level in the imaging conditions can be corrected using the same set of coefficients. Since the selected sensor has an imaging condition with an integration level of 24, images meeting the conditions under this condition are searched again, and the above steps are repeated to correct and update the relative radiometric calibration coefficients. Figure 5 The image shown is a comparison of the image correction results before (left) and after (right) calibration coefficient correction for an image under the imaging condition of 24-level integration series and 3 times gain.

Claims

1. A method for rapid relative radiometric calibration of local stripe noise in optical remote sensing satellite images, characterized in that, The steps include the following: Step 1: For the problem area with stripe noise in the image, find uniformly imaged ground features with uniform reflectivity, so that the uniformly imaged ground feature covers the location of the stripe noise by more than 5 pixels on the left and right, and has a height of more than 1000 rows. Step 2: Perform relative radiometric correction on the image using laboratory or in-orbit relative radiometric calibration coefficients to obtain the corrected image; Step 3: Based on the uniform region selected in Step 1, correct the vertical stripe noise in the corrected image obtained in Step 2, and calculate the scaling coefficient coeff. i The formula is as follows: In the formula, The average gray level of the image over a uniform region. The average grayscale value of the i-th column of the uniform region image, where i is the sensor pixel column number. startcol i represents the leftmost column number of the uniform region. endcol The rightmost column number of the uniform region; Step 4: If the sensor only has a laboratory relative radiometric calibration coefficient, then correct that coefficient: The gain term in the laboratory relative radiation calibration coefficient is extracted and corrected, while the bias term remains unchanged. The formula is as follows: In the formula, Let be the original laboratory relative radiometric calibration coefficient for sensor pixel i. The laboratory relative radiometric calibration coefficient for sensor pixel i after correction; If the sensor already has on-orbit relative radiometric calibration coefficients, then those coefficients should be corrected: The mapping values ​​of each pixel in the on-orbit relative radiometric calibration coefficients at different gray levels are extracted and corrected accordingly, as shown in the following formula: In the formula, Let be the mapped gray value corresponding to the original on-orbit relative radiometric calibration coefficient of sensor pixel i when the gray value is j. Let $n$ be the mapped gray value corresponding to the on-orbit relative radiometric calibration coefficient of sensor pixel $i$ at gray value $j$, and $n$ be the sensor quantization bits. n -1 represents the sensor's saturation grayscale value.

2. The method for rapid relative radiometric calibration of local stripe noise in optical remote sensing satellite images according to claim 1, characterized in that, If the sensor has multiple imaging conditions, change the imaging conditions and repeat steps one to four until the relative radiometric calibration coefficients are corrected and updated under all conditions.

3. The method for rapid relative radiometric calibration of local stripe noise in optical remote sensing satellite images according to claim 2, characterized in that, The imaging conditions are integral series and gain.

4. The method for rapid relative radiometric calibration of local stripe noise in optical remote sensing satellite images according to claim 1, characterized in that, The uniformly imaged ground features are deserts and water bodies.

Images