An airborne photoelectric pod optical axis automatic calibration method

Abstract

The application relates to an airborne photoelectric pod optical axis automatic calibration method, belonging to the photoelectric detection and image technology field, and comprising the following steps: image synchronous acquisition and pretreatment; heterogeneous image feature extraction and robust matching; geometric transformation model solving; optical axis pixel deviation calculation; pixel deviation transmission and electronic axis correction; axis correction result verification and compensation parameter updating. The application can realize rapid and accurate correction of the consistency of visible light and infrared image optical axes.

Description

Technical Field

[0001] This invention relates to the field of photoelectric detection and imaging technology, and in particular to an automatic optical axis calibration method for an airborne photoelectric pod. Background Technology

[0002] With the increasing application of airborne optoelectronic detection systems in modern military reconnaissance, target strike, and civilian telemetry, their core component—the optoelectronic pod—typically integrates multispectral detection methods such as visible light cameras, infrared thermal imagers, and laser rangefinders. To ensure the precise consistency of multi-source detection information in space, that is, for optical payloads of different wavelengths to point to the same target area at any detection distance, optical axis consistency (also known as multi-optical axis parallelism) is a key indicator for evaluating its performance.

[0003] In actual operation, due to factors such as severe vibration, temperature cycling and mechanical stress release in the airborne environment, the optical components inside the electro-optical pod may undergo slight deformation or relative displacement, causing the originally calibrated visible light and infrared optical axes to shift. This results in the visible light image and infrared image being unable to be aligned coaxially, leading to problems such as target line-of-sight deviation, tracking loss, and increased pointing error, which seriously affects the overall combat effectiveness and reliability of the electro-optical pod.

[0004] Currently, the visible light and infrared optical axis calibration of airborne optoelectronic pods mainly falls into two categories: One type is the hardware-assisted calibration method, such as the online calibration system for the relative error of the optical axis of multi-band common aperture optoelectronic equipment disclosed in patent CN114326011A. This system relies on a dedicated calibration plate or laser-assisted device to achieve optical axis alignment. Although it has high accuracy, it has drawbacks such as large equipment size, complex operation, and inability to perform real-time online calibration during flight missions.

[0005] One approach is electronic axis alignment based on heterogeneous image registration, such as the automatic axis alignment multispectral optoelectronic pod method proposed in patent CN202410463182.X, which aligns the infrared field of view center with the visible field of view center by aligning with natural feature targets to achieve electronic axis alignment; and the optoelectronic payload three-axis online calibration system disclosed in patent CN115682835A, which adjusts the infrared optical axis based on the difference in the target's position in visible light and infrared imaging. However, existing heterogeneous image registration and axis alignment schemes still have significant shortcomings: The preprocessing stage is highly generalized but lacks parameter optimization for airborne scenarios: Existing technologies mostly use general filtering and contrast enhancement methods, without designing precise preprocessing schemes such as median filtering combined with CLAHE (block, contrast limit 2.0) to address the noise characteristics of the airborne environment and the differences in modalities of heterogeneous images. This can easily lead to residual noise or excessive local contrast enhancement, affecting the stability of subsequent feature extraction.

[0006] Feature extraction is not adequately adapted to the modal differences between visible light and infrared images: Existing patents mostly use a single feature operator (such as SIFT, ORB or deep learning features) for heterogeneous image matching. They do not design a differentiated feature extraction strategy of "visible light SIFT + infrared improved LBP" to take into account the characteristics of visible light images with rich texture and clear corners, while infrared images have sparse texture and prominent edges. This results in a high mismatch rate and insufficient registration accuracy in heterogeneous image feature matching.

[0007] Insufficient integrity of the alignment link: Some patents only focus on image registration itself (such as CN115601407A and CN117788312A), failing to integrate the registration results with optical axis deviation calculation and real-time compensation to form an end-to-end electronic alignment process. This makes it difficult to meet the dynamic alignment requirements of airborne optoelectronic pods in complex battlefield environments. By calculating pixel deviations through image processing and transmitting the data to the FPGA logic processing module for electronic alignment, rapid and accurate correction of the optical axis consistency between visible light and infrared images can be achieved. Summary of the Invention

[0008] In view of this, the purpose of this invention is to provide an automatic optical axis calibration method for an airborne optoelectronic pod, which calculates pixel deviations through image processing and transmits them to an FPGA logic processing module for electronic axis calibration, thereby achieving rapid and accurate calibration of the optical axis consistency between visible light and infrared images.

[0009] The technical solution adopted by this invention to solve the above-mentioned technical problems is: an automatic optical axis calibration method for an airborne optoelectronic pod, comprising the following steps: S1. Image Synchronous Acquisition and Preprocessing: The visible light camera and infrared thermal imager of the airborne optoelectronic pod are controlled by the airborne main control unit to operate synchronously, acquire visible light images and infrared images under the same ground background scene, and preprocess the visible light images and infrared images. S2. Heterogeneous Image Feature Extraction and Robust Matching: Extract cross-band stable common features from the preprocessed visible light image and infrared image respectively, establish the initial feature matching relationship between the two images, and then use the random sampling consensus algorithm to remove mismatched feature points to obtain high-precision feature matching point pairs. S3. Geometric Transformation Model Solution: Based on high-precision feature matching point pairs, solve the geometric transformation model between visible light images and infrared images, and establish the spatial mapping relationship between the two; S4. Calculation of optical axis pixel deviation: Retrieve the preset coordinates of the visible light image alignment center, and map the coordinates of the visible light image alignment center to the infrared image coordinate system through the solved geometric transformation model to obtain the mapped coordinates. Calculate the horizontal and vertical pixel deviations between the mapped coordinates and the infrared image alignment center. S5. Pixel Deviation Transmission and Electronic Axis Correction: The horizontal and vertical pixel deviations obtained in step S4 are sent to the FPGA logic processing module. The FPGA logic processing module performs pixel-level translation transformation on the real-time video stream of the infrared or visible light image to make the axis correction center points of the infrared and visible light images accurately coincide.

[0010] S6. Verification of Axis Calibration Results and Update of Compensation Parameters: After the electronic axis calibration is completed, the verification process is triggered: Repeat steps S1 to S4 to calculate the corrected pixel deviation; if the pixel deviation is less than or equal to the preset accuracy threshold, the axis calibration is determined to be successful, and the pixel deviation calculated this time is stored as a fixed compensation parameter in the non-volatile memory of the FPGA logic processing module as the reference offset for subsequent real-time image correction; if the pixel deviation is greater than the preset accuracy threshold, steps S2 to S5 are re-executed for iterative correction until the accuracy requirements are met.

[0011] As a preferred embodiment, in step S1, the preprocessing specifically involves: applying median filtering to denoise the visible light image and the infrared image respectively, and then applying an adaptive histogram equalization algorithm to enhance the contrast of the denoised visible light image and the infrared image. This process removes image noise interference, optimizes image quality, and effectively improves the stability and accuracy of subsequent feature extraction, laying the foundation for subsequent image registration. As a preferred embodiment, in step S2, common features that are stable across bands are extracted from the preprocessed visible light image and infrared image, respectively, specifically as follows: For visible light images, the scale-invariant feature transform algorithm is used to extract corner features; For infrared images, an improved local binary mode algorithm is used to extract edge and texture features; As a preferred option, in step S2, a fast nearest neighbor search algorithm is used to establish an initial feature matching relationship between two images.

[0012] As a preferred embodiment, in step S2, the interior point determination threshold of the random sampling consensus algorithm is 3 pixels, the maximum number of iterations is 1000, and at least 30 pairs of high-precision matching points are obtained. This filters out interference data and outliers, resulting in high-precision, high-reliability feature matching point pairs, ensuring the accuracy of subsequent spatial mapping relationship solving.

[0013] As a preferred approach, in step S3, the geometric transformation model is an affine transformation model, and the affine matrix is ​​solved using the least squares method. This eliminates the viewpoint deviation, scale deviation, and positional deviation caused by heterogeneous imaging, achieving accurate registration of infrared and visible light images and providing data support for the quantification of optical axis deviation.

[0014] As a preferred option, in step S4, the coordinates of the visible light image alignment center are mapped to the infrared image coordinate system by solving the affine matrix.

[0015] As a preferred embodiment, in step S5, the FPGA logic processing module performs pixel-level translation transformation on the real-time video stream of the infrared image or visible light image, specifically as follows: If the visible light image is corrected based on the infrared image, the entire visible light image frame is shifted horizontally by the negative value of the horizontal pixel deviation and vertically by the negative value of the vertical pixel deviation. If the infrared image is corrected based on the visible light image, the entire infrared image frame is shifted horizontally by a positive value of the horizontal pixel deviation and vertically by a positive value of the vertical pixel deviation. This process requires no physical adjustment of the pod structure, responds quickly, and can compensate for optical axis drift caused by factors such as temperature and vibration in real time, ensuring that the output image always remains centered.

[0016] As a preferred option, in step S6, the preset accuracy threshold is 2, and the pixel deviation includes horizontal pixel deviation and vertical pixel deviation. When the absolute values ​​of both the horizontal pixel deviation and the vertical pixel deviation are less than or equal to 2, the axis alignment is determined to be successful.

[0017] The beneficial effects of this application are as follows: 1. This application relies entirely on the visible light camera and infrared thermal imager of the airborne optoelectronic pod, without the need to add dedicated calibration equipment such as collimators and target plates, which greatly reduces the system complexity and hardware cost.

[0018] 2. This application directly transmits the calculated horizontal and vertical pixel deviations to the FPGA logic processing module for image translation correction. There is no need to physically adjust the pod servo mechanism. The axis correction process is fully automatic, has a fast response speed, and can compensate for optical axis drift in real time.

[0019] 3. This application uses a scale-invariant feature transform algorithm to extract corner features for visible light images and an improved local binary mode algorithm to extract edge and texture features for infrared images. It also combines a fast nearest neighbor search algorithm to establish an initial feature matching relationship between two images and then uses a random sampling consensus algorithm to remove mismatched feature points. This can maintain high-precision registration in complex environments (such as rain, fog, and changes in lighting) and ensure the stability and reliability of the alignment results.

[0020] 4. This application has steps for verifying the alignment effect and updating the compensation parameters, forming a closed-loop feedback to ensure that the alignment accuracy meets the system index requirements, and dynamically updates the compensation parameters to adapt to optical axis drift during long-term use. Attached Figure Description

[0021] Figure 1 This is a flowchart of the present invention.

[0022] Figure 2 This is a schematic diagram of the visible light channel and mid-wave infrared channel of the airborne optoelectronic pod after optical electronic alignment processing in the distant scene of a village and town according to the present invention.

[0023] Figure 3 This is a schematic diagram of the visible light channel and mid-wave infrared channel of the airborne optoelectronic pod after optical electronic alignment processing in a distant urban high-rise scene according to the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that, in the description of this invention, unless otherwise stated, "a plurality of" means two or more; the terms "upper," "lower," "left," "right," "inner," "outer," "front end," "rear end," "head," "tail," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0025] Please see Figure 1-3 This invention provides an automatic optical axis calibration method for an airborne optoelectronic pod. This embodiment is based on a certain type of airborne optoelectronic pod, which integrates a 1920×1080 resolution visible light camera, a 640×512 resolution infrared thermal imager, and is equipped with an embedded image processing unit. The specific implementation steps are as follows: Step 1: Synchronous image acquisition and preprocessing.

[0026] Before the flight mission, the airborne electro-optical pod is mounted under the aircraft. The main control unit issues a synchronous acquisition command, causing the visible light camera and infrared thermal imager to simultaneously image a region of the ground with rich texture features (such as buildings or road intersections), acquiring visible light and infrared images. Subsequently, the visible light and infrared images are preprocessed separately.

[0027] Specifically, adopt Median filtering templates are used to denoise both visible light and infrared images. Median filtering, a non-linear smoothing technique, sorts the gray values ​​of all pixels in the neighborhood of the target pixel and takes the median as the new gray value for that pixel. It effectively suppresses typical noises such as salt-and-pepper noise and impulse noise, while preserving image edge and texture details to the greatest extent possible. The specific steps are as follows: choose The filter template (i.e., the convolution kernel size is) ), taking the target pixel as the center, covering its 8 adjacent pixels, forming The local window is used to sort the gray values ​​of the 9 pixels within the window by size, and the median value is taken as the new gray value of the target pixel. All pixels in the visible light image and infrared image are traversed to complete the full image denoising, providing a clean image foundation for subsequent feature extraction.

[0028] Specifically, the Adaptive Histogram Equalization (CLAHE) algorithm is used for contrast enhancement. The specific parameters are set as follows: contrast limit amplitude is 2.0, and the image is divided into... Within each sub-block, histogram equalization is performed independently. The specific steps are as follows: Block partitioning: The preprocessed visible light image and infrared image are divided into blocks. For non-overlapping sub-block regions, each sub-block has its histogram calculated independently.

[0029] Contrast Limit: Set the contrast limit amplitude to 2.0. This means that the number of pixels in each sub-block histogram that exceeds this threshold will be cropped, and the cropped portion will be evenly distributed to other gray levels to prevent excessive enhancement of local contrast.

[0030] Interpolation fusion: After equalization of each sub-block, bilinear interpolation is used to stitch and fuse the results, eliminating the boundary effect between sub-blocks and obtaining an enhanced image with a smooth transition.

[0031] This embodiment improves the contrast of texture, edge and other features in heterogeneous images while preserving the overall structure of the image, providing better image quality for SIFT feature extraction and improved stable LBP feature extraction in subsequent steps.

[0032] Step 2: Heterogeneous image feature extraction and robust matching.

[0033] Specifically, for the preprocessed visible light and infrared images, the following steps are used to extract features: For visible light images, the Scale Invariant Feature Transform (SIFT) algorithm is used to extract corner features, and the feature descriptor has a dimension of 128.

[0034] For infrared images, given their relatively limited texture details, an improved Local Binary Pattern (LBP) algorithm is used to extract edge and texture features.

[0035] Subsequently, the Fast Nearest Neighbor Search (FLANN) algorithm is used to establish the initial feature matching relationship between the two images, and then the Random Sample Consensus (RANSAC) algorithm is used to iteratively eliminate outliers. The RANSAC algorithm has an interior point detection threshold of 3 pixels and a maximum number of iterations of 1000, ultimately obtaining no fewer than 30 pairs of high-precision matching points to ensure the stability of the model solution.

[0036] Step 3: Solve the geometric transformation model.

[0037] Specifically, based on the selected high-precision matching point pairs, and combining the geometric relationship of the imaging plane of the airborne pod with the installation attitude parameters, an affine transformation model is selected to describe the spatial mapping relationship between the visible light image and the infrared image. The 3×3 affine matrix H is solved using the least squares method, so that the points in the visible light image coordinate system... Points that can be mapped to the infrared image coordinate system This affine transformation model can effectively compensate for rotation, translation, and small-scale scaling deviations between the two lenses.

[0038] Step 4: Calculate the optical axis pixel deviation.

[0039] Specifically, the preset visible light image alignment center coordinates (960, 540) are retrieved. Using the affine transformation matrix H obtained in step 3, these coordinates are mapped to the infrared image coordinate system, yielding the corresponding mapped coordinates. ; Calculate the mapped coordinates The difference between the horizontal pixel deviation and the standard alignment center of the infrared image (320, 256) is used to obtain the horizontal pixel deviation. Vertical pixel deviation : Horizontal pixel deviation: Vertical pixel deviation: This deviation value quantifies the degree of misalignment between the visible light optical axis and the infrared optical axis at the center of the image under the current state.

[0040] Step 5: Pixel deviation transmission and electronic axis correction.

[0041] horizontal pixel deviation and vertical pixel deviation The data is transmitted in real time to the FPGA logic processing module of the airborne optoelectronic pod via the LVDS high-speed data interface. The FPGA logic processing module has a pre-set image translation correction logic: if the visible light image is corrected based on the infrared image, the entire visible light image frame is shifted horizontally by the negative value of the horizontal pixel deviation and vertically by the negative value of the vertical pixel deviation; if the infrared image is corrected based on the visible light image, the entire infrared image frame is shifted horizontally by the positive value of the horizontal pixel deviation and vertically by the positive value of the vertical pixel deviation. The translated image is output to the display terminal via the FPGA logic processing module. At this point, the center points of the infrared and visible light images are precisely aligned, achieving electronic alignment.

[0042] Step 6: Verify the alignment effect and update the compensation parameters.

[0043] Specifically, after electronic alignment is completed, the system automatically triggers image synchronization acquisition again, repeating steps 1 to 4 to calculate the corrected pixel deviation. The system's preset accuracy threshold is set to 2. Pixel deviation includes horizontal and vertical pixel deviations. Alignment is considered successful when the absolute values ​​of both horizontal and vertical pixel deviations are less than or equal to 2. The calculated pixel deviation is written as a fixed compensation parameter into the non-volatile (Flash) memory of the FPGA logic processing module, and this compensation is automatically added during subsequent real-time image correction. If the pixel deviation is greater than the preset accuracy threshold (i.e., the absolute value of either the horizontal or vertical pixel deviation is greater than 2), steps 2 to 5 are automatically repeated until the pixel deviation meets the requirements or the maximum number of iterations is reached (maximum number of iterations is 5). The final updated compensation parameters are synchronously sent to the target tracking module of the airborne optoelectronic pod to ensure high-precision image registration in subsequent tasks.

[0044] Of course, the present invention is not limited to the embodiments described above. Several other embodiments based on the design concept of the present invention are also provided below.

[0045] For example, in Embodiment 2, unlike Embodiment 1 described above, the feature extraction algorithm used in step 2 is different: this embodiment uses a deep learning feature extraction network, specifically a pre-trained SuperPoint network to extract image feature points, combined with a SuperGlue network for feature matching. This method has stronger robustness in texture-scarce scenes (such as sea surfaces and deserts), and can further improve registration accuracy.

[0046] For example, in Embodiment 3, unlike Embodiment 1 described above, the execution method of electronic axis alignment in step 5 is as follows: After receiving the pixel deviation, the FPGA logic processing module not only performs whole-frame translation correction, but also performs sub-pixel level interpolation translation according to actual needs. The bilinear interpolation algorithm is used to improve the correction accuracy to 0.1 pixel level, which meets the application scenarios of optical axis consistency with higher precision requirements.

[0047] Figure 2 and Figure 3 In the image, the grayscale sub-image on the left represents the mid-infrared channel imaging, while the main image represents the visible light channel imaging. Both images use their own independent pixel coordinate systems, with the top left corner of the image defined as the origin, the x-axis horizontally, and the y-axis vertically. The coordinates marked in the image are the center pixel coordinates of the optical axis after the corresponding channel has been corrected.

[0048] After completing the optical axis calibration in a distant view of a village or town: (e.g.) Figure 2As shown, the mid-wave infrared sub-image coordinates (313, 265) are the center pixel coordinates of the optical axis in the original infrared image's own pixel coordinate system, corresponding to the center of the view of the pavilion structure in the real scene; the visible light main image coordinates (958, 549) are the center pixel coordinates of the optical axis in the original visible light image's own pixel coordinate system, corresponding to the same pavilion structure in the real scene. All ground objects in the scene correspond to the same pixel index position in both the infrared and visible light images, achieving pixel-level view axis alignment in both visible light and mid-wave infrared dual channels.

[0049] After optical axis calibration in a distant urban high-rise scene: (e.g.) Figure 3 As shown, the mid-wave infrared sub-image coordinates (314, 265) are the center pixel coordinates of the optical axis in the original infrared image's own pixel coordinate system, corresponding to the center of the line of sight of the spire of the Chinese-style roof structure; the visible light main image coordinates (960, 547) are the center pixel coordinates of the optical axis in the original visible light image's own pixel coordinate system, corresponding to the same spire of the Chinese-style roof structure in the real scene. All ground objects in the scene correspond to the same pixel index position in both the infrared and visible light images, achieving pixel-level line-of-sight alignment in both visible light and mid-wave infrared dual channels.

[0050] It should be noted that the above embodiments are only used to illustrate the present invention, but the present invention is not limited to the above embodiments. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for automatic optical axis calibration of an airborne optoelectronic pod, characterized in that, Includes the following steps: S1. Image Synchronous Acquisition and Preprocessing: The visible light camera and infrared thermal imager of the airborne optoelectronic pod are controlled by the airborne main control unit to operate synchronously, acquire visible light images and infrared images under the same ground background scene, and preprocess the visible light images and infrared images. S2. Heterogeneous Image Feature Extraction and Robust Matching: Extract cross-band stable common features from the preprocessed visible light image and infrared image respectively, establish the initial feature matching relationship between the two images, and then use the random sampling consensus algorithm to remove mismatched feature points to obtain high-precision feature matching point pairs. S3. Geometric Transformation Model Solution: Based on high-precision feature matching point pairs, solve the geometric transformation model between visible light images and infrared images, and establish the spatial mapping relationship between the two. S4. Calculation of optical axis pixel deviation: Retrieve the preset coordinates of the visible light image alignment center, and map the coordinates of the visible light image alignment center to the infrared image coordinate system through the solved geometric transformation model to obtain the mapped coordinates. Calculate the horizontal and vertical pixel deviations between the mapped coordinates and the infrared image alignment center. S5. Pixel Deviation Transmission and Electronic Axis Correction: The horizontal and vertical pixel deviations obtained in step S4 are sent to the FPGA logic processing module. The FPGA logic processing module performs pixel-level translation transformation on the real-time video stream of the infrared or visible light image to make the axis correction center points of the infrared and visible light images accurately coincide. S6. Verification of alignment results and update of compensation parameters: After the electronic alignment is completed, the verification process is triggered: repeat steps S1 to S4 to calculate the corrected pixel deviation. If the pixel deviation is less than or equal to the preset accuracy threshold, the axis alignment is considered successful, and the pixel deviation calculated this time is stored as a fixed compensation parameter in the non-volatile memory of the FPGA logic processing module as the reference offset for subsequent real-time image correction; if the pixel deviation is greater than the preset accuracy threshold, steps S2 to S5 are repeated for iterative correction until the accuracy requirements are met.

2. The automatic optical axis calibration method for an airborne optoelectronic pod according to claim 1, characterized in that, In step S1, the preprocessing specifically involves: using median filtering to denoise the visible light image and the infrared image respectively, and then using an adaptive histogram equalization algorithm to enhance the contrast of the denoised visible light image and the infrared image.

3. The automatic optical axis calibration method for an airborne optoelectronic pod according to claim 2, characterized in that, In step S2, common features that are stable across bands are extracted from the preprocessed visible light image and infrared image, respectively. Specifically: For visible light images, the scale-invariant feature transform algorithm is used to extract corner features; For infrared images, an improved local binary mode algorithm is used to extract edge and texture features.

4. The automatic optical axis calibration method for an airborne optoelectronic pod according to claim 3, characterized in that, In step S2, the fast nearest neighbor search algorithm is used to establish the initial feature matching relationship between the two images.

5. The automatic optical axis calibration method for an airborne optoelectronic pod according to claim 4, characterized in that, In step S2, the threshold for determining the interior point of the random sampling consensus algorithm is 3 pixels, the maximum number of iterations is 1000, and finally at least 30 pairs of high-precision matching points are obtained.

6. The automatic optical axis calibration method for an airborne optoelectronic pod according to claim 5, characterized in that, In step S3, the geometric transformation model is an affine transformation model, and the affine matrix is ​​solved by the least squares method.

7. The automatic optical axis calibration method for an airborne optoelectronic pod according to claim 6, characterized in that, In step S4, the coordinates of the visible light image alignment center are mapped to the infrared image coordinate system by solving the affine matrix.

8. The automatic optical axis calibration method for an airborne optoelectronic pod according to claim 7, characterized in that, In step S5, the FPGA logic processing module performs pixel-level translation transformation on the real-time video stream of the infrared or visible light image, specifically as follows: If the visible light image is corrected based on the infrared image, the entire visible light image frame is shifted horizontally by the negative value of the horizontal pixel deviation and vertically by the negative value of the vertical pixel deviation. If the infrared image is corrected based on the visible light image, the entire frame of the infrared image is shifted horizontally by the positive value of the horizontal pixel deviation and vertically by the positive value of the vertical pixel deviation.

9. The automatic optical axis calibration method for an airborne optoelectronic pod according to claim 8, characterized in that, In step S6, the preset accuracy threshold is 2. The pixel deviation includes horizontal pixel deviation and vertical pixel deviation. When the absolute values ​​of both the horizontal and vertical pixel deviations are less than or equal to 2, the axis calibration is determined to be successful.

Images