Satellite image positioning precision detection method and device, equipment and storage medium

Abstract

The invention relates to the technical field of satellite data processing, in particular to a satellite image positioning precision detection method, device and equipment and a storage medium, and the method comprises the steps: obtaining a to-be-detected first target satellite image and n first reference high-resolution images matched with the first target satellite image, and carrying out the standardization processing and radiation enhancement of the first target satellite image and the n first reference high-resolution images, obtaining a second target satellite image and n second reference high-resolution images; cutting the second target satellite image into n third target satellite images corresponding to the n second reference high-resolution images; converting the second reference high-resolution image into a third reference high-resolution image which is the same as a third target satellite image corresponding to the second reference high-resolution image in image size and resolution; and calculating the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image by adopting a preset optical flow algorithm, and taking the pixel offset as a positioning precision result. According to the technical scheme, the positioning precision of the satellite image can be detected with high efficiency and high precision.

Description

Technical Field

[0001] The present invention relates to the field of satellite data processing technology, and in particular to a method, device, equipment and storage medium for detecting positioning accuracy of satellite images. Background Art

[0002] The positioning of satellite images refers to the determination of the geographical location corresponding to the satellite images observed by the satellite. The higher the positioning accuracy of the satellite images, the more accurate the real geographical location corresponding to the satellite images, and the more effective the subsequent application of satellite images in various fields. Therefore, the positioning accuracy of satellite images is one of the important indicators of satellite images. The traditional positioning accuracy test method mainly extracts and matches the satellite images to be tested with the reference images with geographical location tags in the database manually, and determines the positioning accuracy of the satellite images to be tested based on the matching results. This process is relatively cumbersome, not only in terms of accuracy, but also in terms of processing efficiency and automation. Summary of the invention

[0003] In order to solve the problems in the related art, the embodiments of the present disclosure provide a method, device, equipment and storage medium for detecting the positioning accuracy of satellite images.

[0004] In a first aspect, an embodiment of the present disclosure provides a method for detecting positioning accuracy of satellite images, including:

[0005] Acquire a first target satellite image to be detected and acquire n first reference high-resolution images matching the first target satellite image, where n is an integer greater than or equal to 1;

[0006] The first target satellite image and the n first reference high-resolution images are both subjected to standardization processing and radiation enhancement to obtain a second target satellite image and n second reference high-resolution images;

[0007] According to the position range of the n second reference high-resolution images, the second target satellite image is cropped into n third target satellite images corresponding to the positions of the n second reference high-resolution images;

[0008] For each second reference high-resolution image and a corresponding third target satellite image thereof, converting the second reference high-resolution image into a third reference high-resolution image having the same image size and resolution as the corresponding third target satellite image;

[0009] A preset optical flow algorithm is used to calculate the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image as a positioning accuracy result.

[0010] In a possible implementation manner, the step of acquiring a satellite image of a first target to be detected includes:

[0011] Obtain the original satellite image to be detected and its image channel and calibration coefficient;

[0012] Based on the calibration coefficient and the image channel, calibration calculation is performed on the original satellite image to be detected to obtain a target satellite image;

[0013] Cloud detection is performed on the target satellite image, and the target satellite image whose cloud occlusion ratio exceeds a predetermined threshold is used as the first target satellite image to be detected.

[0014] In a possible implementation, the acquiring n first reference high-resolution images matching the first target satellite image includes:

[0015] Based on the first target satellite image, searching from a pre-stored high-resolution image library for n high-resolution images taken by a predetermined satellite, located within the position range of the first target satellite image, having the same image channel as the first target satellite image, having the highest image accuracy level, and being taken at a time closest to that of the first target satellite image as reference high-resolution images;

[0016] The high-resolution image library records a plurality of high-resolution images and their corresponding shooting satellites, shooting times, location ranges, image channels, and positioning accuracy levels of the images.

[0017] In a possible implementation, the method further includes:

[0018] The high-resolution images are downloaded periodically to update the high-resolution image library.

[0019] In a possible implementation, the using of an optical flow algorithm to calculate the pixel offset from the third reference high-resolution image to the third target satellite image as a positioning accuracy result includes:

[0020] If the resolution of the third target satellite image is greater than or equal to the preset resolution, an optical flow algorithm is used to calculate the pixel offset from the third reference high-resolution image to the third target satellite image as a positioning accuracy result;

[0021] The method further comprises:

[0022] If the resolution of the third target satellite image is less than the preset resolution, the third target satellite image and the third reference high-resolution image are processed to obtain a fourth reference high-resolution image to the fourth target satellite image;

[0023] An optical flow algorithm is used to calculate the pixel offset from the fourth reference high-resolution image to the fourth target satellite image as a positioning accuracy result.

[0024] In a possible implementation, the using an optical flow algorithm to calculate the pixel offset from the fourth reference high-resolution image to the fourth target satellite image as a positioning accuracy result includes:

[0025] The fourth reference high-resolution image is used as the starting image, and the fourth target satellite image is used as the intermediate image and the ending image, and is input into a pre-trained optical flow calculation model, and the optical flow calculation model is executed to obtain a forward optical flow vector from the fourth reference high-resolution image to the fourth target satellite image. and the backward optical flow vector from the fourth target satellite image to the fourth target satellite image

[0026] Based on the forward optical flow vector and the backward optical flow vector Calculate the final optical flow vector F = (F u , F v ) as the positioning accuracy result;

[0027] in, is the lateral optical flow in the forward optical flow vector, is the longitudinal optical flow in the forward optical flow vector, is the lateral optical flow in the backward optical flow vector, is the longitudinal optical flow in the backward optical flow vector, and m is the resolution multiple of the fourth target satellite image relative to the third target satellite image.

[0028] In a possible implementation, the method further includes:

[0029] Obtaining a pixel offset of a preset control point set from the pixel offsets of each pixel;

[0030] Based on the pixel offset of the control point set, a random sampling consistency detection algorithm is used to remove abnormal points in the control point set, and the pixel offset of normal control points is obtained as the final positioning accuracy result.

[0031] In a second aspect, an embodiment of the present disclosure provides a device for detecting positioning accuracy of satellite images, including:

[0032] An image acquisition module is configured to acquire a first target satellite image to be detected and acquire n first reference high-resolution images matching the first target satellite image, where n is an integer greater than or equal to 1;

[0033] A preprocessing module is configured to perform standardization processing and radiation enhancement on the first target satellite image and the n first reference high-resolution images to obtain a second target satellite image and n second reference high-resolution images;

[0034] A cropping module is configured to crop the second target satellite image into n third target satellite images corresponding to the positions of the n second reference high-resolution images according to the position range of the n second reference high-resolution images;

[0035] A conversion module is configured to convert, for each second reference high-resolution image and a corresponding third target satellite image, the second reference high-resolution image into a third reference high-resolution image having the same image size and resolution as the corresponding third target satellite image;

[0036] The positioning analysis module is configured to use a preset optical flow algorithm to calculate the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image as a positioning accuracy result.

[0037] In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising a memory and a processor, wherein the memory is used to store one or more computer instructions, and wherein the one or more computer instructions are executed by the processor to implement a method as described in any one of the first aspects.

[0038] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium on which computer instructions are stored. When the computer instructions are executed by a processor, a method as described in any one of the first aspects is implemented.

[0039] The technical solution provided by the embodiments of the present disclosure has the following technical effects:

[0040] 1. Significantly improve detection accuracy

[0041] The present disclosure adopts high-resolution images as high-precision reference data. High-resolution images not only have extremely high spatial resolution, but also are rich in ground object information. The rich details of these high-resolution images provide more reference points for positioning analysis, and provide an accurate and reliable data basis for subsequent operations such as feature matching and positioning analysis; at the same time, combined with the feature matching-based optical flow method, it can more accurately capture the subtle feature changes between the reference image and the image to be detected, and also has strong dynamic tracking capabilities, and can track the changes of these feature points in real time, even when there are scale differences or slight deformations between images. It can maintain a high matching accuracy, thereby improving the accuracy of the positioning detection results.

[0042] In addition, the traditional image library used for reference has a long update cycle, which makes it difficult to capture and reflect the real-time changes of landmarks, thus limiting its timeliness and accuracy. In contrast, the high-resolution image library of the present invention can be updated regularly, realizing a near real-time update mechanism, which can closely track the passage of time and subtle or significant changes in the appearance of landmarks. This dynamic update feature ensures that the present invention can match the most appropriate and accurate reference high-resolution image for the satellite image to be detected. Its immediacy and sensitivity provide crucial support for the precise positioning analysis of satellite images, greatly improving the user experience and information value.

[0043] Moreover, the present invention innovatively adopts an optical flow algorithm to calculate the pixel offset of each pixel from a reference high-resolution image to the satellite image to be detected as a positioning accuracy result, which can accurately capture the position offset between the reference image and the image to be detected, thereby ensuring the accuracy of the positioning detection results.

[0044] 2. Significantly improved processing efficiency

[0045] The present invention automates the entire positioning accuracy analysis process from data processing to feature matching, especially the introduction of the optical flow algorithm, which greatly reduces the need for manual intervention and improves processing efficiency. At the same time, automated processing also reduces the possibility of human error and improves the accuracy and reliability of positioning detection.

[0046] 3. Strong adaptability and generalization capabilities

[0047] The optical flow algorithm used in the present disclosure can automatically learn and optimize parameters from a large amount of data through a self-supervised learning mechanism without the need for manual labeling, and therefore has strong adaptability and generalization capabilities.

[0048] 4. Excellent robustness and reliability

[0049] The optical flow algorithm used in the present disclosure estimates the optical flow by combining the sequence information of multiple frames of images. This spatiotemporal analysis method performs well in dealing with complex situations such as occlusion and motion blur. In addition, the robustness of the deep learning algorithm also enables the model to maintain stable performance in data with noise or outliers.

[0050] 5. Wide applicability

[0051] The present disclosure has extremely high flexibility in the selection of high-resolution images for reference, which is mainly reflected in its wide applicability. Specifically, the present disclosure not only uses high-resolution images of Landsat series satellites as reference data sources, showing excellent performance, but can also be seamlessly extended to the application of other high-resolution satellite data sources. This flexibility stems from the versatility of the design of the present disclosure and its compatibility with multiple data formats.

[0052] The present disclosure also shows strong adaptability to the satellite images to be tested. Regardless of which satellite system these satellite images to be tested come from, what resolution they have, and what sensor they are acquired with, as long as they contain sufficient ground object information and feature points, they can be effectively evaluated for positioning accuracy using the method disclosed in the present disclosure. This wide applicability is due to the use of feature matching optical flow method, which can automatically extract and match key features from complex and changeable image data, thereby achieving high-precision positioning analysis.

[0053] In addition, the wide applicability of the present disclosure is also reflected in its scalability. With the continuous development of remote sensing technology and the continuous emergence of new satellite systems, the method provided by the present disclosure can quickly adapt to new data sources and new application scenarios through simple adjustments and optimizations. This capability makes the present disclosure have great potential and value in the future field of remote sensing data processing and analysis.

[0054] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Other features, objectives and advantages of the present disclosure will become more apparent through the following detailed description of non-limiting embodiments in conjunction with the accompanying drawings. In the accompanying drawings:

[0056] Figure 1 A flowchart of a method for detecting positioning accuracy of satellite images provided by an embodiment of the present disclosure is shown.

[0057] Figure 2 A structural block diagram of a satellite image positioning accuracy detection device according to an embodiment of the present disclosure is shown.

[0058] Figure 3 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0059] Figure 4 A schematic diagram showing the structure of a computer system suitable for implementing the method of the embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0060] Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement them. In addition, for the sake of clarity, parts not related to the description of the exemplary embodiments are omitted in the accompanying drawings.

[0061] In the present disclosure, it should be understood that terms such as "include" or "have" are intended to indicate the presence of features, numbers, steps, behaviors, components, parts, or a combination thereof disclosed in the present specification, and are not intended to exclude the possibility that one or more other features, numbers, steps, behaviors, components, parts, or a combination thereof exist or are added.

[0062] It should also be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0063] The following glossary is explained before the embodiments:

[0064] High-resolution imagery: high-resolution satellite imagery;

[0065] Pixel: the smallest unit that makes up a satellite image;

[0066] Image channel: used to indicate the color or wavelength of a satellite image, such as infrared channel, shortwave channel, visible light channel, etc.

[0067] Figure 1 FIG. 1 is a flow chart showing a method for detecting the positioning accuracy of a satellite image provided by an embodiment of the present disclosure. Figure 1 As shown, the satellite image positioning accuracy detection method includes the following steps S101-S105:

[0068] In step S101, a first target satellite image to be detected is obtained and n first reference high-resolution images matching the first target satellite image are obtained.

[0069] Here, the first target satellite image to be detected can be taken by any camera sensor carried by any satellite system, such as a geostationary satellite, a polar-orbiting satellite, etc. There is no restriction on the resolution, size, image channel, etc. of the first target satellite image. As long as the satellite image is taken by the camera sensor on the satellite, the positioning accuracy can be detected using the method provided in the present invention.

[0070] Here, the first reference high-resolution image refers to a high-resolution satellite image used as a reference for evaluating the positioning accuracy of the first target satellite image. Usually, the resolution of the first target satellite image is low, and one frame of the first target satellite image usually matches multiple frames of the first reference high-resolution image. The matching here means that the position range of the continuous area formed by multiple first reference high-resolution images is the same as the position range of the first target satellite image and the image channel of the first reference high-resolution image is the same as that of the first target satellite image. Of course, in order to better evaluate the first target satellite image, the shooting time of the first reference high-resolution image and the first target satellite image is preferably close, and as a reference image for positioning accuracy evaluation, the positioning accuracy of the first reference high-resolution image is higher.

[0071] In step S102, the first target satellite image and the n first reference high-resolution images are both subjected to standardization processing and radiation enhancement to obtain a second target satellite image and n second reference high-resolution images.

[0072] Here, in order to ensure the consistency of the first target satellite image and the first reference high-resolution image, the first target satellite image and the n first reference high-resolution images can be standardized. The standardization process refers to scaling the pixel values ​​of the image proportionally, removing the unit restriction of the pixel values, and converting them into dimensionless pure values, so that the first target satellite images and the first reference high-resolution images of different magnitudes can be compared.

[0073] Here, in order to increase the image contrast of the image data and improve the visual effect of the image, after the standardization process, the first target satellite image and the n first reference high-resolution images can also be subjected to radiation enhancement (Radiance Enhancement), so as to obtain the second target satellite image and n second reference high-resolution images.

[0074] In step S103, according to the position range of the n second reference high-resolution images, the second target satellite image is cropped into n third target satellite images corresponding to the positions of the n second reference high-resolution images;

[0075] Here, in order to ensure the spatial consistency and comparability of the second target satellite image and the second reference high-resolution image from different sources, and to provide an accurate and reliable data basis for subsequent operations such as feature matching and positioning analysis, n second reference high-resolution images can be projected into the second target satellite image to obtain n projection areas of the n second reference high-resolution images in the second target satellite image, and then the second target satellite image is cropped into n third target satellite images corresponding to the positions of the n second reference high-resolution images according to the n projection areas, and the n second reference high-resolution images correspond one-to-one with the n third target satellite images, and a same-name point association is established. It should be noted here that the position range of each second reference high-resolution image can be searched in the second target satellite image according to the position range of the n second reference high-resolution images, such as the longitude and latitude range, corresponding to the pixel row and column number range in the second target satellite image, so that the n projection areas of the n second reference high-resolution images in the second target satellite image can be obtained.

[0076] In step S104, for each second reference high-resolution image and its corresponding third target satellite image, the second reference high-resolution image is converted into a third reference high-resolution image having the same image size and resolution as the corresponding third target satellite image.

[0077] Here, the second reference high-resolution image is a high-resolution satellite image, which is usually higher in resolution than the corresponding third target satellite image. In order to better compare, resampling can be performed to convert the second reference high-resolution image into a third reference high-resolution image with the same image size and resolution as the corresponding third target satellite image.

[0078] In step S105, a preset optical flow algorithm is used to calculate the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image as a positioning accuracy result.

[0079] Here, the optical flow algorithm is a computer vision technology used to estimate the pixel motion between consecutive image frames. The concept of optical flow is based on the following assumption: in a short period of time, the movement of objects will cause changes in the pixel positions in the image, and these changes can be captured by analyzing the image sequence. Therefore, the present disclosure innovatively adopts the optical flow algorithm to calculate the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image as the positioning accuracy result, where the pixel offset can include the lateral offset and longitudinal offset of each pixel.

[0080] The optical flow algorithm is innovatively used to calculate the pixel offset of each pixel from the reference high-resolution image to the satellite image to be detected as the positioning accuracy result. It can accurately capture the position offset between the reference image and the image to be detected, and ensure the accuracy of the positioning detection results.

[0081] In a possible implementation manner, the step of acquiring a satellite image of a first target to be detected includes:

[0082] Obtain the original satellite image to be detected and its image channel and calibration coefficient;

[0083] Based on the calibration coefficient and the image channel, calibration calculation is performed on the original satellite image to be detected to obtain a target satellite image;

[0084] Cloud detection is performed on the target satellite image, and the target satellite image whose cloud occlusion ratio exceeds a predetermined threshold is used as the first target satellite image to be detected.

[0085] Here, calibration coefficients are a set of parameters used to convert the raw satellite image data captured by the sensor on the satellite into actual surface reflectivity or emissivity. These coefficients reflect the characteristics of the sensor and its sensitivity to light of different wavelengths. By calibrating the raw satellite image, errors caused by factors such as the sensor's own characteristics, changes in the solar altitude angle, and atmospheric influences can be corrected, thereby obtaining more realistic surface physical quantities.

[0086] For example, for the original satellite image of the visible light channel taken by the polar-orbiting satellite, the visible light reflectance can be calculated based on the original brightness value DN1 of the pixel of the original satellite image and the calibration coefficients k0, k1 and k2 of the visible light channel. The calculation formula is: value1 = k0 + k1 × DN1 + k2 × DN1 2 Based on the calculated visible light reflectance of each pixel, the target satellite image can be obtained.

[0087] Or, for example, for the original satellite image of the infrared channel taken by the polar-orbiting satellite, the infrared channel radiance can be calculated based on the original brightness value DN2 of the pixel of the original satellite image and the calibration coefficients k3, k4 and k5 of the infrared channel. The calculation formula is: value2 = k3 + k4 × DN2 + k5 × DN2 2 , based on the calculated infrared channel radiance of each pixel, the target satellite image can be obtained.

[0088] Or, for example, the visible reflectance and infrared channel radiance of the geostationary satellite may also be obtained through a lookup table.

[0089] Here, the obtained target satellite image can also be quickly detected by cloud. If it is detected that the cloud obstruction does not exceed a predetermined threshold value such as 50%, it means that the target satellite image is less obstructed by clouds, and positioning accuracy detection can be performed, and the subsequent detection process can continue; if it is detected that the cloud obstruction exceeds a predetermined threshold value such as 50%, it means that the target satellite image is more obstructed by clouds and is not suitable for positioning accuracy detection. The process ends. At this time, other original satellite images taken by the same sensor on the same satellite can be re-acquired to perform calibration calculations and cloud detection to obtain the first target satellite image to be detected.

[0090] In a possible implementation, the acquiring n first reference high-resolution images matching the target satellite image includes:

[0091] Based on the target satellite image, searching from a pre-stored high-resolution image library for n high-resolution images taken by a predetermined satellite, located within the position range of the target satellite image, taken at a time closest to the target satellite image, with the same image channel as the target satellite image and with the highest image accuracy level as reference high-resolution images;

[0092] The high-resolution image library records a plurality of high-resolution images and their corresponding shooting satellites, shooting times, location ranges, image channels, and positioning accuracy levels of the images.

[0093] Here, the high-resolution image library stores a large number of high-resolution images, which refer to high-resolution images. These high-resolution images can be Landsat (Landsat) series (such as Landsat8, Landsat 9) high-resolution images or other high-resolution satellite images. The high-resolution image is a satellite image that has been calibrated and calculated, and can be full-channel TIFF (Tagged Image File Format) image data.

[0094] Here, the high-resolution image can be a high-resolution image of the area where the corresponding landmark point is located. The landmark point can be a natural lake area with significant landmark features, such as Qinghai Lake, Nam Co Lake, Selin Co Lake and Hulun Lake. These areas not only have distinct landmark features, but also cover a wide range of altitudes and diverse geographical environments. They are ideal test areas and provide rich and challenging test scenarios for the effectiveness and accuracy verification of the detection method provided by the present disclosure. It should be noted here that the first target satellite image to be detected is also a satellite image of the area where the above-mentioned landmark point is located.

[0095] Here, the high-resolution image is used as a reference to detect the positioning accuracy of other satellite images. Therefore, in order to ensure the accuracy of the detection, high-resolution images with a cloud occlusion ratio less than a predetermined ratio, such as 20%, can be selected to construct the high-resolution image library.

[0096] Here, for the convenience of management and application, each high-resolution image in the high-resolution image library records the corresponding shooting satellite, shooting time, location range (such as latitude and longitude range), image channel (such as infrared channel, visible light channel, etc.), and image positioning accuracy level (level set according to predetermined rules) of the high-resolution image. In this way, after obtaining the first target satellite image to be detected, n high-resolution images taken by a predetermined satellite, located within the location range of the first target satellite image, with the same image channel as the first target satellite image, the highest image accuracy level, and the closest shooting time to the first target satellite image can be searched from the pre-stored high-resolution image library as reference high-resolution images.

[0097] It should be noted that the predetermined satellite may be set by the inspector. The high-resolution image located within the position range of the first target satellite image may be determined according to the position range of each high-resolution image.

[0098] The high-resolution images in this embodiment not only have extremely high spatial resolution, which can clearly show the surface details, but also have excellent temporal resolution and spectral resolution, which can provide multi-dimensional and all-round surface information. These rich high-resolution image data sources not only broaden the scope of remote sensing applications, but also provide solid and reliable data support for the realization of the high-precision positioning analysis disclosed in this disclosure.

[0099] In a possible implementation, the method further includes:

[0100] The high-resolution images are downloaded periodically to update the high-resolution image library.

[0101] In this implementation, the high-resolution images in the high-resolution impact library can be manually downloaded from the official. In order to ensure the timely update of the high-resolution images in the high-resolution image library, they can also be downloaded from the official at a scheduled time. For example, full-channel TIFF image data of areas with cloud occlusion <20% and landmark points can be downloaded once a day.

[0102] This implementation method updates the high-resolution image library by periodically downloading high-resolution images to ensure the freshness of the high-resolution images in the high-resolution image library. This ensures that the first reference high-resolution image that matches the first target satellite image and has the most recent shooting time can be found in the high-resolution image library, providing crucial support for the precise positioning analysis of satellite images and making subsequent positioning detection results more accurate.

[0103] In a possible implementation, the using of an optical flow algorithm to calculate the pixel offset from the third reference high-resolution image to the third target satellite image as a positioning accuracy result includes:

[0104] If the resolution of the third target satellite image is greater than a preset threshold, an optical flow algorithm is used to calculate the pixel offset from the third reference high-resolution image to the third target satellite image as a positioning accuracy result;

[0105] The method further comprises:

[0106] If the resolution of the third target satellite image is less than a preset threshold, the third target satellite image and the third reference high-resolution image are processed to obtain a fourth reference high-resolution image to the fourth target satellite image;

[0107] An optical flow algorithm is used to calculate the pixel offset from the fourth reference high-resolution image to the fourth target satellite image as a positioning accuracy result.

[0108] Here, in order to ensure the consistency of the reference high-resolution image and its corresponding target satellite image, it is necessary to convert the high-resolution second reference high-resolution image into a third reference high-resolution image with the same image size and resolution as the corresponding third target satellite image. If the resolution of the third target satellite image is large, greater than or equal to the preset resolution (such as less than or equal to 500 meters), it can be ensured that the third reference high-resolution image and the third target satellite image have higher spatial details and clearer feature boundaries. At this time, the optical flow algorithm can be directly used to calculate the pixel offset from the third reference high-resolution image to the third target satellite image as the positioning accuracy result. If the resolution of the third target satellite image is small, less than the preset resolution (such as greater than 500 meters), in order to improve the clarity of the geographic feature boundaries in the image and enhance the accuracy of subsequent optical flow matching, it is necessary to perform resolution up-processing on it. For example, a linear interpolation algorithm can be used to perform resolution up-processing on the third reference high-resolution image and its corresponding third target satellite image, and their resolution is doubled (or other specified multiples). The linear interpolation method estimates the value of the new pixel by considering the value of the surrounding pixels, thereby improving the resolution of the image without introducing too much human error. The images after the resolution upgrade (the fourth reference high-resolution image and the fourth target satellite image) will have higher spatial details and clearer feature boundaries, which will be beneficial for subsequent operations such as feature matching and positioning analysis.

[0109] In a possible implementation, the using an optical flow algorithm to calculate the pixel offset from the fourth reference high-resolution image to the fourth target satellite image as a positioning accuracy result includes:

[0110] The fourth reference high-resolution image is used as the starting image, and the fourth target satellite image is used as the intermediate image and the ending image, and is input into a pre-trained optical flow calculation model, and the optical flow calculation model is executed to obtain a forward optical flow vector from the fourth reference high-resolution image to the fourth target satellite image. and the backward optical flow vector from the fourth target satellite image to the fourth target satellite image

[0111] Based on the forward optical flow vector and the backward optical flow vector Calculate the final optical flow vector F = (F u , F v ) as the positioning accuracy result;

[0112] Here, the optical flow calculation model is an optical flow calculation model pre-trained by a self-supervised optical flow learning method. The optical flow calculation model can indirectly evaluate the change in positioning accuracy by analyzing the changes in surface features between the target satellite image and the high-resolution image, and obtain the positioning accuracy of the target satellite image with reference to the positioning accuracy of the high-resolution image. For example, the optical flow calculation model can be a SelFlow optical flow calculation model.

[0113] Here, the input of the optical flow calculation model is three frames of images, representing the starting image, the middle image and the ending image respectively. In this embodiment, the fourth reference high-resolution image can be used as the starting image, and the fourth target satellite image can be used as the middle image and the ending image, and input into the optical flow calculation model to calculate the optical flow, so that the forward optical flow from the fourth reference high-resolution image to the fourth target satellite image can be obtained. and a backward optical flow from the fourth target satellite image to the fourth target satellite image in, is the lateral optical flow in the forward optical flow vector, is the longitudinal optical flow in the forward optical flow vector, is the lateral optical flow in the backward optical flow vector, is the longitudinal optical flow in the backward optical flow vector.

[0114] Since the intermediate image and the end image for calculating the backward optical flow vector in this embodiment are the same image (both are the fourth target satellite image), the backward optical flow vector can be regarded as the calculation error of the optical flow calculation model (the backward optical flow value is generally 1e -1 ~1e -4 In order to eliminate the influence of model error, the forward optical flow vector is subtracted from the backward optical flow vector as the final optical flow vector. The specific formula is as follows: The purpose of dividing by m is to convert the pixel deviation after the resolution increase back to the pixel deviation before the resolution increase, where m is the resolution multiple of the fourth target satellite image relative to the third target satellite image. In this way, the lateral offset F from the fourth reference high-resolution image pixel to the fourth target satellite image pixel is obtained. u With longitudinal offset F v That is, the final optical flow vector, and the positioning accuracy detected by this implementation is sub-pixel level.

[0115] Here, the set of final optical flow vectors from the n fourth reference high-resolution images to each pixel of the corresponding n fourth target satellite images can be used as the positioning accuracy result.

[0116] It should be noted here that, for the case where the resolution is not increased, the above solution can be referred to, and the third reference high-resolution image is used as the starting image, and the third target satellite image is used as the intermediate image and the ending image, and is input into the optical flow calculation model to calculate the optical flow. In this way, the forward optical flow from the third reference high-resolution image to the third target satellite image will be obtained. and a backward optical flow from the third target satellite image to the third target satellite image in, is the lateral optical flow in the forward optical flow vector, is the longitudinal optical flow in the forward optical flow vector, is the lateral optical flow in the backward optical flow vector, is the longitudinal optical flow in the backward optical flow vector; the forward optical flow vector minus the backward optical flow vector is taken as the final optical flow vector. The specific formula is as follows: Since no resolution increase is performed, there is no need to divide by m as in the above solution.

[0117] In a possible implementation, the method further includes:

[0118] Obtaining a pixel offset of a preset control point set from the pixel offsets of each pixel;

[0119] Based on the pixel offset of the control point set, a random sampling consistency detection algorithm is used to remove abnormal points in the control point set, and the pixel offset of normal control points is obtained as the final positioning accuracy result.

[0120] Here, in order to ensure the accuracy and credibility of the positioning accuracy results, the pixel offset of a reliable control point set (such as a pixel point set on the coastline, etc.) can be obtained from the pixel offset of each pixel to further control the quality of the positioning accuracy results. Then, the random sampling consistency detection method (RANSAC) is used to eliminate false outliers, such as pixels whose pixel offsets are significantly different from those of other pixel offsets. Finally, the pixel offset of the normal control point is obtained as the final positioning accuracy result.

[0121] It should be noted here that when outputting the positioning accuracy result or the final positioning accuracy result, it can be output in the form of a file in a specified format. The positioning accuracy result or the final positioning accuracy result may include the image space coordinates, object space coordinates, sub-pixel level pixel offset and other information of the pixel point, which is convenient for users to view and use.

[0122] The present disclosure also provides a satellite image positioning accuracy detection device, Figure 2The structure block diagram of the satellite image positioning accuracy detection device according to the embodiment of the present disclosure is shown, and the device can be implemented as part or all of the electronic device through software, hardware or a combination of both. Figure 2 As shown, the satellite image positioning accuracy detection device includes:

[0123] The image acquisition module 201 is configured to acquire a first target satellite image to be detected and acquire n first reference high-resolution images matching the first target satellite image, where n is an integer greater than or equal to 1;

[0124] The preprocessing module 202 is configured to perform standardization processing and radiation enhancement on the first target satellite image and the n first reference high-resolution images to obtain a second target satellite image and n second reference high-resolution images;

[0125] The cropping module 203 is configured to crop the second target satellite image into n third target satellite images corresponding to the positions of the n second reference high-resolution images according to the position range of the n second reference high-resolution images;

[0126] The conversion module 204 is configured to convert each second reference high-resolution image and a corresponding third target satellite image thereof into a third reference high-resolution image having the same image size and resolution as the corresponding third target satellite image;

[0127] The positioning analysis module 205 is configured to use a preset optical flow algorithm to calculate the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image as a positioning accuracy result.

[0128] In a possible implementation manner, the part of the image acquisition module that acquires the first target satellite image to be detected is configured as follows:

[0129] Obtain the original satellite image to be detected and its image channel and calibration coefficient;

[0130] Based on the calibration coefficient and the image channel, calibration calculation is performed on the original satellite image to be detected to obtain a target satellite image;

[0131] Cloud detection is performed on the target satellite image, and the target satellite image whose cloud occlusion ratio exceeds a predetermined threshold is used as the first target satellite image to be detected.

[0132] In a possible implementation manner, the part of the image acquisition module that acquires n first reference high-resolution images that match the first target satellite image is configured as follows:

[0133] Based on the first target satellite image, searching from a pre-stored high-resolution image library for n high-resolution images taken by a predetermined satellite, located within the position range of the first target satellite image, having the same image channel as the first target satellite image, having the highest image accuracy level, and being taken at a time closest to that of the first target satellite image as reference high-resolution images;

[0134] The high-resolution image library records a plurality of high-resolution images and their corresponding shooting satellites, shooting times, location ranges, image channels, and positioning accuracy levels of the images.

[0135] In a possible implementation, the device further includes:

[0136] The download module is configured to periodically download high-resolution images to update the high-resolution image library.

[0137] In a possible implementation, the positioning analysis module is configured as follows:

[0138] If the resolution of the third target satellite image is greater than or equal to the preset resolution, an optical flow algorithm is used to calculate the pixel offset from the third reference high-resolution image to the third target satellite image as a positioning accuracy result;

[0139] The device also includes:

[0140] The upscaling module is configured to perform upscaling processing on the third target satellite image and the third reference high-resolution image to obtain a fourth reference high-resolution image to the fourth target satellite image if the resolution of the third target satellite image is less than a preset resolution;

[0141] The calculation module is configured to use an optical flow algorithm to calculate the pixel offset from the fourth reference high-resolution image to the fourth target satellite image as a positioning accuracy result.

[0142] In a possible implementation, the calculation module is configured as follows:

[0143] The fourth reference high-resolution image is used as the starting image, and the fourth target satellite image is used as the intermediate image and the ending image, and is input into a pre-trained optical flow calculation model, and the optical flow calculation model is executed to obtain a forward optical flow vector from the fourth reference high-resolution image to the fourth target satellite image. and the backward optical flow vector from the fourth target satellite image to the fourth target satellite image

[0144] Based on the forward optical flow vector and the backward optical flow vector Calculate the final optical flow vector F = (F u , Fv ) as the positioning accuracy result;

[0145] in, is the lateral optical flow in the forward optical flow vector, is the longitudinal optical flow in the forward optical flow vector, is the lateral optical flow in the backward optical flow vector, is the longitudinal optical flow in the backward optical flow vector, and m is the resolution multiple of the fourth target satellite image relative to the third target satellite image.

[0146] In a possible implementation, the device further includes:

[0147] The quality control module is configured to obtain the pixel offset of a preset control point set from the pixel offset of each pixel; based on the pixel offset of the control point set, a random sampling consistency detection algorithm is used to eliminate abnormal points in the control point set, and the pixel offset of the normal control point is obtained as the final positioning accuracy result.

[0148] The technical terms and technical features mentioned in the implementation manner of the present device are the same as or similar to those mentioned in the implementation manner of the above method. For the explanation and description of the technical terms and technical features involved in the present device, reference may be made to the explanation and description of the implementation manner of the above method, and they will not be repeated here.

[0149] The present disclosure also discloses an electronic device, Figure 3 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0150] like Figure 3 As shown, the electronic device 300 includes a memory 301 and a processor 302, wherein the memory 301 is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor 302 to implement the method according to an embodiment of the present disclosure.

[0151] Figure 4 A schematic diagram showing the structure of a computer system suitable for implementing the method of the embodiment of the present disclosure is shown.

[0152] like Figure 4 As shown, the computer system 400 includes a processing unit 401, which can perform various processes in the above-mentioned embodiments according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage part 408 into a random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the computer system 400 are also stored. The processing unit 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0153] The following components are connected to the I / O interface 405: an input section 406 including a keyboard, a mouse, etc.; an output section 407 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. The drive 410 is also connected to the I / O interface 405 as needed. Removable media 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., are installed on the drive 410 as needed, so that the computer program read therefrom is installed into the storage section 408 as needed. Among them, the processing unit 401 can be implemented as a processing unit such as a CPU, a GPU, a TPU, an FPGA, an NPU, etc.

[0154] In particular, according to an embodiment of the present disclosure, the method described above can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes computer instructions, and the computer instructions are executed by a processor to implement the method steps described above. In such an embodiment, the computer program product can be downloaded and installed from a network through the communication part 409, and / or installed from a removable medium 411.

[0155] It should be noted here that Figure 4 The computer system shown is only an example of a computer system. The computer system may also be of other structures, for example, without the above-mentioned driver, etc., as long as it can implement the method of the embodiment of the present disclosure, and no limitation is made here.

[0156] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0157] The units or modules involved in the embodiments described in the present disclosure may be implemented by software or programmable hardware. The units or modules described may also be set in a processor, and the names of these units or modules do not constitute limitations on the units or modules themselves in some cases.

[0158] As another aspect, the present disclosure further provides a computer-readable storage medium, which may be a computer-readable storage medium included in the electronic device or computer system in the above embodiment; or a computer-readable storage medium that exists independently and is not assembled into a device. The computer-readable storage medium stores one or more programs, and the programs are used by one or more processors to execute the method described in the present disclosure.

[0159] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present disclosure is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other.

Claims

1. A method for detecting positioning accuracy of satellite images, characterized in that: include: Acquire a first target satellite image to be detected and acquire n first reference high-resolution images matching the first target satellite image, where n is an integer greater than or equal to 1; The first target satellite image and the n first reference high-resolution images are both subjected to standardization processing and radiation enhancement to obtain a second target satellite image and n second reference high-resolution images; According to the position range of the n second reference high-resolution images, the second target satellite image is cropped into n third target satellite images corresponding to the positions of the n second reference high-resolution images; For each second reference high-resolution image and a corresponding third target satellite image thereof, converting the second reference high-resolution image into a third reference high-resolution image having the same image size and resolution as the corresponding third target satellite image; A preset optical flow algorithm is used to calculate the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image as a positioning accuracy result.

2. The method according to claim 1, characterized in that The step of acquiring a first target satellite image to be detected includes: Obtain the original satellite image to be detected and its image channel and calibration coefficient; Based on the calibration coefficient and the image channel, calibration calculation is performed on the original satellite image to be detected to obtain a target satellite image; Cloud detection is performed on the target satellite image, and the target satellite image whose cloud occlusion ratio exceeds a predetermined threshold is used as the first target satellite image to be detected.

3. The method according to claim 1, characterized in that: The step of acquiring n first reference high-resolution images matching the first target satellite image comprises: Based on the first target satellite image, searching from a pre-stored high-resolution image library for n high-resolution images taken by a predetermined satellite, located within the position range of the first target satellite image, having the same image channel as the first target satellite image, having the highest image accuracy level, and being taken at a time closest to that of the first target satellite image as reference high-resolution images; The high-resolution image library records a plurality of high-resolution images and their corresponding shooting satellites, shooting times, location ranges, image channels, and positioning accuracy levels of the images.

4. The method according to claim 3, characterized in that The method further comprises: The high-resolution images are downloaded periodically to update the high-resolution image library.

5. The method according to claim 1, characterized in that The method of using a preset optical flow algorithm to calculate the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image as a positioning accuracy result includes: If the resolution of the third target satellite image is greater than or equal to the preset resolution, an optical flow algorithm is used to calculate the pixel offset from the third reference high-resolution image to the third target satellite image as a positioning accuracy result; The method further comprises: If the resolution of the third target satellite image is less than the preset resolution, the third target satellite image and the third reference high-resolution image are processed to obtain a fourth reference high-resolution image to the fourth target satellite image; A preset optical flow algorithm is used to calculate the pixel offset from the fourth reference high-resolution image to the fourth target satellite image as a positioning accuracy result.

6. The method according to claim 5, characterized in that The using a preset optical flow algorithm to calculate the pixel offset from the fourth reference high-resolution image to the fourth target satellite image as a positioning accuracy result includes: The fourth reference high-resolution image is used as the starting image, and the fourth target satellite image is used as the intermediate image and the ending image, and is input into a pre-trained optical flow calculation model, and the optical flow calculation model is executed to obtain a forward optical flow vector from the fourth reference high-resolution image to the fourth target satellite image. and the backward optical flow vector from the fourth target satellite image to the fourth target satellite image Based on the forward optical flow vector and the backward optical flow vector Calculate the final optical flow vector F = (F u , F v ) as the positioning accuracy result; in, is the lateral optical flow in the forward optical flow vector, is the longitudinal optical flow in the forward optical flow vector, is the lateral optical flow in the backward optical flow vector, is the longitudinal optical flow in the backward optical flow vector, and m is the resolution multiple of the fourth target satellite image relative to the third target satellite image.

7. The method according to claim 1, characterized in that The method further comprises: Obtaining a pixel offset of a preset control point set from the pixel offsets of each pixel; Based on the pixel offset of the control point set, a random sampling consistency detection algorithm is used to remove abnormal points in the control point set, and the pixel offset of normal control points is obtained as the final positioning accuracy result.

8. A satellite image positioning accuracy detection device, characterized in that: include: An image acquisition module is configured to acquire a first target satellite image to be detected and acquire n first reference high-resolution images matching the first target satellite image, where n is an integer greater than or equal to 1; A preprocessing module is configured to perform standardization processing and radiation enhancement on the first target satellite image and the n first reference high-resolution images to obtain a second target satellite image and n second reference high-resolution images; A cropping module is configured to crop the second target satellite image into n third target satellite images corresponding to the positions of the n second reference high-resolution images according to the position range of the n second reference high-resolution images; A conversion module is configured to convert, for each second reference high-resolution image and a corresponding third target satellite image, the second reference high-resolution image into a third reference high-resolution image having the same image size and resolution as the corresponding third target satellite image; The positioning analysis module is configured to use a preset optical flow algorithm to calculate the pixel offset of each pixel from the third reference high-resolution image to the third target satellite image as a positioning accuracy result.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that: Computer instructions are stored thereon, and when the computer instructions are executed by a processor, the method steps described in any one of claims 1 to 7 are implemented.

Images