Processing method for extracting line-scan digital camera yaw mode data twill angle

Abstract

The invention belongs to the technical field of camera imaging, and discloses a processing method for extracting a line-scan digital camera yaw mode data twill angle, in a yaw mode, because all pixels of a CCD image the same ground object, relative radiometric calibration can be carried out based on yaw data, however, the yaw data twill angle is influenced by the camera linear array speed and frame frequency, and the linear array camera yaw mode data twill angle is more accurate. The yaw twill angle of each time is not fixed, and the influence on the subsequent calculation of the calibration coefficient is very large, so that the rapid and accurate extraction of the yaw data twill angle becomes a new demand. Aiming at the problems of low timeliness, high computer resource occupation, low calculation precision and the like of an existing yaw data twill angle calculation algorithm, the invention provides a processing method for extracting a linear array camera yaw mode data twill angle, and the processing method comprises the steps of reading yaw data, constructing and calculating an optimal standard deviation operator at each angle, and calculating wave bands one by one. Therefore, the method for efficiently and accurately extracting the data sloping grain angle in the yaw mode of the camera is realized. And reliable angle parameters are provided for subsequent yaw data processing. And a solution thought is provided for subsequent projects with the same requirements.

Description

Technical Field

[0001] The present application belongs to the field of camera imaging technology and discloses a processing method for extracting the twill angle of yaw mode data of a linear array camera. Background Art

[0002] During the long-term operation of satellite-borne sensors, their images will have radiation non-uniformity problems of varying degrees and types. This is due to the non-uniformity of the spatial response of the optical system (such as the different optical transmittances of different parts of the lens), the impossibility of uniform width of the grating slits everywhere, the differences in the responses of various detectors, damage to the detectors, the dark current of the CCD, the inconsistency of the electronic links outside the CCD array, and other problems. These lead to the non-uniformity of infrared images, which seriously affects the imaging quality of infrared sensors.

[0003] Relative radiation correction is a commonly used remote sensing image processing method, which aims to eliminate image radiation differences caused by factors such as uneven surface reflectivity, atmospheric scattering and absorption. This correction method can improve the quality of remote sensing images and make them more consistent with the radiation characteristics of actual objects. In view of the above problems, relative radiation correction can be used to eliminate the radiation inhomogeneity effect caused by inconsistent responses of sensors or charge-coupled device detectors. In ground system processing, relative radiation correction uses a linear correction model.

[0004] DN F inal=a N onuniformity×DN DE +b N onuniformity

[0005] Where DN Final is the final output DN value after radiation correction, DN DE is the DN value after correction in the previous step, a Nonuniformity and b Nonuniformity It is the non-uniformity correction factor, which is also related to the load working status and needs to be monitored and updated regularly.

[0006] Yaw calibration is a method for extracting relative radiation correction coefficients based on ground imaging data statistics. In yaw mode, since all CCD pixels can image the same ground object, relative radiation correction can be performed based on yaw data.

[0007] The normal push-broom imaging and yaw-broom imaging process is as follows Figure 2As shown. In the figure, the inherent non-uniformity of the sensor is represented by different grayscales of each detector element 1 to 5. The image obtained by normal scanning contains both the non-uniformity of the sensor and the differences between different objects, and the two are difficult to separate. If the sensor is rotated 90° so that the spatial dimension of the detector element is parallel to the scanning direction, all detector elements 1 to 5 can scan the C\H\M\R\W objects in turn. At this time, the image obtained does not contain the non-uniformity of the object itself, but shows the inherent non-uniformity of the sensor. This is the principle that yaw data can be used for relative radiation correction.

[0008] The first step in yaw data processing is to align the objects (also called normalization), that is, to make the same objects in the image in the same row, such as Figure 3 Therefore, calculating the accurate yaw twill angle is very important for the yaw calibration process.

[0009] The yaw calibration data has the characteristics of large data volume and inconsistent twill angles for each yaw calibration. Theoretically, the twill angle of each normal yaw data under a fixed speed-to-height ratio is a standard fixed value. However, for satellite-borne hyperspectral payloads, each external transformation, such as the influence of the earth's rotation, changes in sampling frequency, and changes in speed-to-height ratio, will cause the twill angle of each yaw data to be inconsistent. If the theoretical twill angle is used for ground object alignment, the above error will be introduced into the coefficient calculation, resulting in the inability to perform relative radiation correction on hyperspectral images. In order to eliminate this error, it is necessary to calculate the high-precision twill angle of each yaw data as much as possible. With the widespread use of hyperspectral cameras in various industries, push-broom hyperspectral camera yaw calibration has become a stable and reliable calibration mode.

[0010] In yaw mode, since all CCD pixels image the same object, relative radiation calibration can be performed based on the yaw data. However, the yaw data twill angle is affected by the camera linear array speed and frame rate. Each yaw twill angle is not fixed, which has a great impact on the subsequent calibration coefficient. Therefore, fast and accurate extraction of the yaw data twill angle has become a new requirement.

[0011] In his paper “Relative Radiometric Calibration of 90° Yaw of the Coastal Zone Imager of Ocean No. 1D Satellite”, Chen Ru from Wuhan University proposed an automatic standardization method for yaw data based on the detector adjacency relationship, targeting the structural characteristics of the HY-1D satellite coastal zone imager. This method effectively improved the problem of nonlinear distortion of the yaw image of the Ocean No. 1 coastal zone imager and ensured the theoretical premise that the linear array charge coupled device (CCD) detectors can image the same ground object. The camera relative radiometric calibration system was calculated using the 90° yaw radiometric calibration data of the HY-1D coastal imager, and a relative radiometric calibration experiment was carried out on the normal push-scan images of the satellite. After relative radiometric calibration using the yaw radiometric calibration data, the maximum fringe coefficient of all sensor detectors in land scenes was better than 0.33%, the average fringe coefficient was better than 0.04%, and the median fringe coefficient was better than 0.03%; the maximum fringe coefficient of all sensor detectors in ocean scenes was better than 0.48%, the average fringe coefficient was better than 0.07%, and the median fringe coefficient was better than 0.06%. All indicators are better than the on-orbit statistical method based on a large amount of data statistics.

[0012] Liang Deyin from the Remote Sensing Satellite General Department of the China Academy of Space Technology proposed an on-orbit yaw calibration method for hyperspectral imagers based on a 90° yaw maneuver of the satellite platform in the paper "On-orbit yaw calibration method for satellite-borne hyperspectral imagers", including the acquisition of yaw data and data processing and analysis methods. This method was verified based on the on-orbit data of the Resource-1 02D satellite. The results showed that this method has a good effect on the relative radiation correction of the hyperspectral imager. The relative radiation calibration accuracy of all pixels of the detector can reach 1%, and the stripe noise in the original image is well removed. The method proposed in the article can realize high-frequency and full-field relative radiation correction of hyperspectral imagers, laying a foundation for the subsequent quantitative application of hyperspectral data.

[0013] Chinese patent CN113570523B discloses a method for automatically generating and extrapolating relative radiation correction coefficients for optical images, the steps of which include: using a camera on a satellite to collect optical images, parsing the auxiliary data in the yaw calibration data transmitted by the satellite, obtaining the current gear gain Gcur and the current integral series Icur of the camera on the satellite, and setting the target gear gain Gtgt and the target integral series Itgt of the camera on the satellite; performing oblique line correction on the yaw image data; calculating the grayscale histogram of each detector and the comprehensive histogram of all detectors, matching the histograms, and obtaining the relative radiation correction coefficient of the current gear of the camera; performing linear fitting on the relative radiation correction coefficient to obtain the linear radiation correction coefficient; and calculating the relative radiation correction coefficient of the target gear of the camera on the satellite according to the linear extrapolation formula. This method does not rely on manual image selection, can effectively reduce the calibration workload, and the generated relative radiation calibration coefficient has high accuracy, which improves the calibration coefficient quality.

[0014] The algorithms commonly used in the prior art for extracting yaw twill angles are LSD and RANSAC:

[0015] Chinese patent CN117474874A discloses an automated optical remote sensing satellite on-orbit image quality monitoring method and a system in which the edge feature area selection process of ground objects is carried out using an LSD algorithm and a threshold-based feature set screening, and effective edge feature screening is performed based on the edge signal-to-noise ratio.

[0016] Chinese patent CN106909902A, a remote sensing target detection method based on an improved hierarchical saliency model, mentions extracting the prior features of airport and oil tank targets from the input image to be detected from top to bottom, and generating target feature maps respectively. For the airport target, LSD (Line Segment Detector) is used to perform line detection on the input image to be detected to obtain the line detection result.

[0017] Chinese patent CN116563349A discloses a hyperspectral image processing method, device, electronic device and medium, which mentions that the SURF algorithm is used to search for registration points from the first hyperspectral image; the LSD algorithm is used to obtain the registration line pair from the first hyperspectral image, and the sub-region division is performed according to the registration line pair, wherein the sub-region division includes the original line segment obtained by the LSD algorithm, and the neighborhood of the original line segment is divided into the left neighborhood and the right neighborhood according to the gradient direction. In the neighborhood of the straight line of each line segment, when the distance from any pixel to the straight line and the distance to the vertical bisector meet the preset value, the straight line is used as the registration line. The RANSAC method is used to refine the registration points, construct the homography matrix vector variable, and when the homography matrix vector variable is kept minimum, the homography matrix of the transformation between the images to be matched is calculated, and the homography matrix is ​​calculated and the minimum value is obtained by the SVD algorithm, wherein the image to be matched is the first hyperspectral image corresponding to the registration point and the registration line.

[0018] Chinese patent CN114972625A discloses a method for generating a hyperspectral point cloud based on RGB spectral super-resolution technology, which mentions calculating the epipolar geometry for each image matching pair, optimizing and improving the matching pair through the RANSAC algorithm, and chaining the feature points in the matching pair to obtain the initialization image pair with the strongest reconstruction and the least outliers.

[0019] Chinese patent CN114218857A discloses a hyperspectral computational imaging method, system and related equipment for multi-spectral acquisition, which mentions calculating the projection transformation matrix between the projection image and other spectral feature images according to the corresponding points. In the embodiment of the present invention, 7 projection transformation matrices can be obtained. At the same time, in the process of obtaining the projection transformation matrix, the RANSAC algorithm (random sampling consensus algorithm) is used to process the projection transformation matrix to ensure the robustness of the projection transformation matrix.

[0020] The LSD algorithm is robust to noise and can process images with noise and complex details, but the computational complexity is high when processing large images. RANSAC can be used in combination with other algorithms (such as Hough transform) to improve the effect of straight line detection, but it is necessary to set appropriate parameters (such as the number of iterations and thresholds). Improper selection may affect the accuracy of the results. When processing yaw data, these algorithms expose problems such as low timeliness, high computer resource usage, low computational accuracy, and unsuitability for automated processes. Summary of the invention

[0021] In view of the above problems, the present invention proposes a processing method for extracting the twill angle of the yaw mode data of a linear array camera, including reading the yaw data, building and calculating the optimal standard deviation operator at each angle, performing calculations band by band, and obtaining the optimal angle, so as to realize a method for efficiently and accurately extracting the twill angle of the data under the yaw mode of the camera. It provides reliable angle parameters for the subsequent processing of the yaw data. It also provides a solution for subsequent projects with the same requirements. The contents of the present invention are as follows:

[0022] The present invention provides a processing method for extracting the twill angle of yaw pattern data of a linear array camera, and the technical point of the method is that the method specifically comprises the following steps:

[0023] Step 1: read the yaw data header file information, obtain the basic information of the image, including the number of bands, the number of pixels, and the total number of frames, read the image and determine whether the data is valid;

[0024] Step 2, read the valid data in step 1 into the cache to obtain a single-band image;

[0025] Step 3, randomly select N pixel numbers and M frame numbers for the single-band image obtained in step 2;

[0026] Step 4: Set the cycle start and end angle β st and β end ;

[0027] Step 5: loop through the start and end angles β in step 4 st and β endThe value between , the angle is β;

[0028] Step 6, loop through the M frame numbers in step 3, the frame number is m;

[0029] Step 7, through the point (1, m), β is K, establish a straight line equation, the intersection array of this equation and the column where the N pixel numbers are located is recorded as array I, through the formula Computes the standard deviation of array I, where is the average value of array I;

[0030] Step 8: For each angle, a standard deviation array II with a length of M can be obtained, and the average of array II is

[0031] Step 9: Set the stop angle β in step 4 st and β end All angles between Find the minimum value of β best ;

[0032] Step 10, Beta best is the twill angle of this set of data.

[0033] In order to better implement the above technical solution, in the processing method for extracting the twill angle of the yaw pattern data of a linear array camera of the present invention, the number of bands Bands, the number of pixels Samples and the total number of frames Lines in step 1 satisfy Liens>2*Samples.

[0034] In order to better implement the above technical solution, the N pixel numbers and M frame numbers in step 3 of the processing method for extracting the twill angle of the yaw pattern data of the linear array camera of the present invention should satisfy: Samples<M<Liens-Samples, 1<N<Samples.

[0035] In order to better implement the above technical solution, the angle β in step 4 of the processing method for extracting the twill angle of the yaw pattern data of the linear array camera of the present invention is st and β end Should satisfy: -90 / 180*pi<β st <β end <90 / 180*pi.

[0036] Compared with the prior art, the processing method for extracting the twill angle of the yaw pattern data of a linear array camera of the present invention has the following beneficial effects:

[0037] The processing method for extracting the twill angle of the yaw mode data of the linear array camera of the present invention can quickly and accurately extract the twill angle of the yaw data without manual parameter adjustment, and converts the semi-automatic calibration process into a fully automatic process. In addition, the algorithm occupies very few resources, has high timeliness and accuracy, and provides a reliable calibration coefficient for the subsequent relative radiation calibration of hyperspectral data, and also provides a solution for subsequent projects with the same requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0039] Figure 1 It is a flow chart of a processing method for extracting the twill angle of the yaw mode data of a linear array camera according to the present invention;

[0040] Figure 2 A schematic diagram of the yaw data principle of a processing method for extracting the twill angle of the yaw mode data of a linear array camera according to the present invention;

[0041] Figure 3 It is a schematic diagram of a yaw image after alignment according to a processing method for extracting twill angles of yaw mode data of a linear array camera of the present invention;

[0042] Figure 4 A schematic diagram of yaw data of a processing method for extracting twill angles of yaw mode data of a linear array camera according to the present invention;

[0043] Figure 5 A schematic diagram of a twill angle comparison of a processing method for extracting twill angles of yaw pattern data of a linear array camera according to the present invention;

[0044] Figure 6 The original data of three yaw calibrations of a certain load. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0046] The technical solutions provided by various embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0047] Example

[0048] A processing method for extracting the twill angle of yaw pattern data of a linear array camera specifically comprises the following steps:

[0049] Step 1: Read the yaw data header file information, obtain the basic information of the image, including the number of bands, the number of pixels, and the total number of frames. Read the image and determine whether the data is valid. Yaw calibration requires a large amount of data for statistical calculation. When the total number of frames is too small, the accuracy of the calibration coefficient cannot be guaranteed. Figure 4 As shown, Samples = 512, Lines = 23000, Bands = 76, and Lines is much larger than Samples.

[0050] Step 2, read the valid data in step 1 into the cache to obtain a single-band image;

[0051] Step 3: Randomly select N pixel numbers and M frame numbers for the single-band image obtained in step 2. N and M should satisfy: Samples<M<Liens-Samples, 1<N<Samples. M=200, N=100;

[0052] Step 4: Set the cycle start and end angle β st and β end .

[0053] Among them, β st =-60 / 180*pi, β end =60 / 180*pi.

[0054] Step 5: loop through the start and end angles β in step 4 st and β end The value between .

[0055] The step size is set to 0.01 (unit radian).

[0056] Step 6, loop through the frame number selected in step 3, the frame number is m;

[0057] Step 7, through the point (1, m), β is K, establish a straight line equation, the intersection array of this equation and the column where the N pixel numbers are located is recorded as array I, through the formula Computes the standard deviation of array I, where is the average value of array I;

[0058] Step 8: For each angle, a standard deviation array II with a length of M can be obtained, and the average of array II is

[0059] Step 9: Change the starting and ending angles of step 4 to β st and β end All angles between Find the minimum value;

[0060] like Figure 5 As shown, the angle numbered 27 is the twill angle of the image.

[0061] Step 10, by formula β best =β st +(num-1)*0.01, we can get the twill angle of this group of data. best =0.7872. Figure 4 The right side is the corrected image.

[0062] Comparative Example

[0063] Figure 6 The raw data of three yaw calibrations of a certain load are shown. It can be seen that the twill angle of the data obtained for each yaw is not consistent. The twill angle of the first yaw has been calculated to be 0.787198 radians. Based on this angle, the raw data of the first yaw is normalized, and the result is shown in Figure d. The normalized data is basically aligned and can be used for the next step. Then, we also normalized the second and third yaw data according to 0.787198 radians, and the results are shown in Figures e and f. It can be seen that the subsequent data is directly regulated based on the twill angle calculated for the first yaw. Since their twill angles are not consistent, it can be seen that there is still twill after regulation, which means that the twill angle needs to be recalculated for each yaw. Without an accurate and fast automatic twill angle calculation algorithm, it will be difficult to achieve automatic yaw calibration. The inability to fix the twill angle is one of the main difficulties in automatic yaw calibration.

[0064] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A processing method for extracting the twill angle of yaw pattern data of a linear array camera, characterized in that: The following steps are involved: Step 1, read the header file information of the yaw data, obtain the basic information of the yaw data, including the number of bands, the number of pixels, and the total number of frames, read the yaw data and determine whether the data is valid; Step 2: If the yaw data in step 1 is valid, read it into the cache to obtain single-band data; Step 3, randomly select N pixel numbers and M frame numbers for the single-band data obtained in step 2; Step 4: Set the cycle start and end angle β st and β end ; Step 5: loop through the start and end angles β in step 4 st and β end The value between , the angle is β; Step 6, loop through the M frame numbers in step 3, the frame number is m; Step 7, through the point (1, m), β is K, establish a straight line equation, the intersection array of this equation and the column where the N pixel numbers are located is recorded as array I, through the formula Compute the standard deviation of array I, where is the average value of array I; Step 8: For each angle, a standard deviation array II with a length of M can be obtained. The average of array II is Step 9: Set the stop angle β in step 4 st and β end All angles between Find the minimum value of β best ; Step 10, Beta best is the twill angle of this set of data.

2. A processing method for extracting the twill angle of the yaw pattern data of a linear array camera according to claim 1, characterized in that: In the step 1, the number of bands Bands, the number of pixels Samples and the total number of frames Lines satisfy Liens>2*Samples.

3. The method for extracting the twill angle of the yaw pattern data of a linear array camera according to claim 1, characterized in that: In step 3, the N pixel numbers and the M frame numbers should satisfy: Samples<M<Liens-Samples, 1<N<Samples.

4. The method for extracting the twill angle of the yaw pattern data of a linear array camera according to claim 1, characterized in that: The angle β in step 4 st and β end Should satisfy: -90 / 180*pi<β st <β end <90 / 180*pi.

Images