Geographic positioning precision analysis method for medium-resolution satellite image

Through the combination of Gaussian filter remapping, SIFT and RANSAC algorithms, the problem of remote sensing image geolocation accuracy dependent on coastline is solved, and high-precision automated evaluation of geolocation accuracy of central resolution satellite images is realized, thereby improving the positioning of coastline-free areas Accuracy evaluation capability.

CN118982572AActive Publication Date: 2024-11-19NAT SATELLITE METEOROLOGICAL CENT
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202411071803.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-19
Estimated Expiration
2044-08-06

AI Technical Summary

Technical Problem

In the prior art, the geolocation accuracy of remote sensing images mainly depends on the coastline, and it is impossible to effectively evaluate the geolocation accuracy of areas without coastlines.

Method used

Using the Gaussian filter-based true image remapping, SIFT feature point matching algorithm and RANSAC algorithm methods, the geolocation accuracy of center-resolution satellite images is automatically evaluated, and outliers are eliminated, which improves the positioning accuracy.

Benefits of technology

It realizes high-precision automated evaluation of geolocation accuracy of satellite remote sensing images, solves the problem of dependent on coastline positioning accuracy, and improves the positioning accuracy evaluation ability in coastline-free areas.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982572A_ABST
    Figure CN118982572A_ABST
Patent Text Reader

Abstract

The invention provides a geographic positioning precision analysis method for a medium-resolution satellite image, and relates to the technical field of geographic positioning, and the method comprises the steps: obtaining a control point image corresponding to a true value image and geographic position information; acquiring a satellite image reference channel image and geographical location information; segmenting the satellite image by using an Otsu threshold method to obtain a cloud region image and a non-cloud region image; remapping the true value image by using Gaussian filtering; extracting cloud region and non-cloud region images and feature points of the remapped images, and performing feature matching by using an SIFT feature point matching algorithm; performing outlier matching point pair elimination by using an RANSAC algorithm according to the matching point pairs, extracting position information of accurate matching point pairs according to true value image geographic position information and satellite image geographic position information, and converting the position information into positioning errors; and carrying out positioning precision inspection according to the positioning error to obtain a positioning precision inspection result. The technical problem that the current positioning precision mainly depends on the coastline is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of geographic positioning technology, and in particular to a method for analyzing the geographic positioning accuracy of medium-resolution satellite images. Background Art

[0002] Remote sensing images are the basis of remote sensing and many applications. As spatial data, remote sensing images have the concept of spatial geographic location. No matter what the application problem is, it is necessary to determine the correspondence between image information and ground position. The geographic positioning of satellite images is an important prerequisite for satellite data preprocessing, and it is very important to be able to obtain geographic positioning accuracy.

[0003] The geolocation of remote sensing images is the process of calculating the coordinates of images observed by satellite-borne remote sensing instruments in the ground-based coordinate system. The existing technical methods for evaluating the geolocation accuracy of remote sensing images mainly determine the geolocation error of the satellite to be detected based on the remote sensing image data, sea and land masks and benchmark coastline landmark points of the satellite to be detected. The coastline identification method relies on the accuracy of sea and land template data. At the same time, the number of landmark points that can be used for identification is limited, and it is impossible to evaluate the geolocation accuracy of areas without coastlines. Therefore, there is an urgent need to develop an image geolocation accuracy evaluation method that does not rely on coastlines. Summary of the invention

[0004] The present invention provides a method for analyzing the geographic positioning accuracy of medium-resolution satellite images, which solves the technical problem that the current positioning accuracy mainly depends on the coastline.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] In a first aspect, a method for analyzing the geolocation accuracy of medium-resolution satellite images is provided, the method comprising:

[0007] Obtain the control point image corresponding to the true value image and the geographical location information of the true value image;

[0008] Obtain satellite image reference channel images and satellite image geographic location information;

[0009] The satellite image is segmented using the Otsu threshold method to obtain images of cloud area and non-cloud area;

[0010] Remap the true value image using Gaussian filtering to obtain a remapped image;

[0011] Extract feature points of cloud area and non-cloud area images and remapped images;

[0012] Use SIFT feature point matching algorithm to perform feature matching to obtain matching point pairs;

[0013] According to the matching point pairs, the RANSAC algorithm is used to remove outlier matching point pairs to obtain accurate matching point pairs;

[0014] According to the geographic location information of the true value image and the satellite image, the location information of the exact matching point pairs is extracted and converted into the positioning error;

[0015] The positioning accuracy is tested based on the positioning error to obtain the positioning accuracy test result.

[0016] Furthermore, the control point image corresponding to the true value image and the geographical location information of the true value image are obtained from the high-precision control point image library, including:

[0017] Get the ground truth image from the high-precision control point image library:

[0018] According to the high-precision control point image library, the corresponding control point image of the true value image is retrieved to obtain the corresponding control point image of the value image;

[0019] The geographic location information is extracted from the true value image and the corresponding control point image to obtain the true value image geographic location information.

[0020] Furthermore, the satellite image reference channel image and satellite image geographic location information are obtained from the load database to be evaluated, including:

[0021] Obtain satellite images from the database of loads to be assessed;

[0022] According to the load database to be evaluated, the reference channel image of the satellite image is retrieved to obtain the reference channel image of the satellite image;

[0023] The geographic location information is extracted from the satellite image reference channel image to obtain the satellite image geographic location information.

[0024] Furthermore, the satellite image is segmented using the Otsu threshold method to obtain images of cloud areas and non-cloud areas, including:

[0025] Obtain satellite image benchmark channel images;

[0026] Convert satellite image base channel images into grayscale images;

[0027] Calculate the histogram of the gray level of each pixel in the grayscale image to obtain the number of pixels at each gray level in the image;

[0028] The number of pixels at each gray level is divided by the total number of pixels to obtain the pixel probability distribution at each gray level;

[0029] Using the Otsu threshold method, set the threshold to segment the grayscale image into two categories, and calculate the variance between the two categories after segmentation to obtain the between-class variance.

[0030] Between-class variance = w0×w1×(μ0-μ1) 2

[0031] Among them, w0 and w1 are the sum of the pixel probabilities of category 0 and category 1, respectively, and μ0 and μ1 are the average grayscale values ​​of category 0 and category 1, respectively;

[0032] According to the inter-class variance, the threshold that maximizes the inter-class variance is selected as the optimal threshold;

[0033] According to the optimal threshold, the grayscale image is binarized, and the pixel grayscale level greater than the threshold is set as the cloud area, and the pixel grayscale level less than or equal to the threshold is set as the non-cloud area.

[0034] Further, the true value image is remapped using Gaussian filtering to obtain a remapped image, including:

[0035] Get the control point image corresponding to the true value image;

[0036] According to the control point image corresponding to the true value image, read the longitude and latitude range of the true value image, and find the corresponding area in the longitude and latitude coordinate map of the corresponding load;

[0037] In the satellite payload corresponding area, find the corresponding coordinate position on the true value image pixel by pixel;

[0038] According to each corresponding coordinate on the true value image, Gaussian filtering is used to extract the pixel mean, and the pixel mean is projected to the coordinate of the corresponding position of the satellite payload to obtain the remapped image.

[0039] Furthermore, feature points of the cloud area and non-cloud area images and the remapped images are extracted, including:

[0040] Obtain images of cloud area and non-cloud area, and remapped images;

[0041] Based on the cloud area and non-cloud area images and the remapped images, the Gaussian difference function is used at different scales to detect the interest points in the image that are invariant to scale and rotation, so as to obtain the extreme points;

[0042] Based on the extreme points, the location and scale of the key points are determined by fitting a refined model;

[0043] Based on the local gradient direction of the image, a main direction is assigned to each key point;

[0044] In the neighborhood around the key point, the local gradient of the image is measured at the selected scale, and the descriptor of the key point is generated by calculating the local gradient of the image to obtain the feature points of the cloud area and non-cloud area images and the remapped image.

[0045] Furthermore, the SIFT feature point matching algorithm is used to perform feature matching to obtain matching point pairs, including:

[0046] Obtain the location, scale, main direction and descriptor of feature points of cloud area and non-cloud area images and remapped images;

[0047] Calculate the similarity between each feature point of the cloud area and non-cloud area images and all feature points of the remapped image;

[0048] According to the similarity threshold, feature points are screened to obtain matching point pairs.

[0049] Furthermore, based on the matching point pairs, the RANSAC algorithm is used to remove outlier matching point pairs to obtain accurate matching point pairs, including:

[0050] Get matching point pairs and set them as data sets;

[0051] Randomly select the minimum required number of data points from the dataset as the initial model;

[0052] Based on the initial model, the parameters of the model are estimated using the selected data points to obtain an estimated model;

[0053] Calculate the fitting error between the remaining matching point pairs in the data set and the estimated model;

[0054] According to the preset threshold, the points whose fitting error with the estimated model is less than the threshold are marked as inliers, and the points whose fitting error is greater than the threshold are marked as outliers;

[0055] Compare the number of inliers for each estimated model and select the model with the largest number of inliers as the best model;

[0056] Repeat the above steps until the preset maximum number of iterations is reached or the number of inliers meets the preset conditions, and select the model with the largest number of inliers as the final exact matching model;

[0057] Use the final exact matching model to remove outliers and obtain accurate matching point pairs.

[0058] Furthermore, based on the geographic location information of the true image and the satellite image, the location information of the exact matching point pairs is extracted and converted into positioning errors, including:

[0059] Obtain the precise matching points between the true image and the satellite image, and obtain the latitude and longitude position information of the matching points;

[0060] According to the latitude and longitude position information of the matching point, find four grid points in the neighborhood of the matching point and obtain the latitude and longitude coordinates of the four grid points;

[0061] According to the latitude and longitude position information of the matching point and the latitude and longitude grid of the payload image, the position of the matching point in sub-pixel coordinates is calculated;

[0062] According to the sub-pixel coordinates of the matching points, the elevation information of the matching points in the sub-pixel coordinates is fitted;

[0063] According to the sub-pixel coordinates of the matching points, the longitude and latitude coordinates of the matching points in the sub-pixel coordinates are calculated;

[0064] The grid coordinate position of the matching point is subtracted from the position in the true value image to obtain the positioning error in the cross-track and along-track directions.

[0065] Further, a positioning accuracy test is performed according to the positioning error to obtain a positioning accuracy test result, including:

[0066] Obtain the positioning errors in the cross-track and along-track directions, and calculate the statistics of the positioning errors;

[0067] According to the statistics of positioning error, the positioning accuracy test result is obtained.

[0068] The above solution of the present invention includes at least the following beneficial effects:

[0069] The above-mentioned scheme of the present invention, in order to solve the problem that the current positioning accuracy mainly depends on the coastline, uses true value image reprojection based on Gaussian filter for remapping, uses SIFT algorithm for feature point matching, and uses RANSAC algorithm for outlier removal, thereby realizing an automated evaluation method for the geographic positioning accuracy of satellite remote sensing images based on a high-precision control point image library. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a flow chart of a method for analyzing the geographic positioning accuracy of medium-resolution satellite images provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0071] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0072] like Figure 1 As shown, an embodiment of the present invention provides a method for analyzing the geographic positioning accuracy of medium-resolution satellite images, the method comprising:

[0073] Step 11, obtaining the control point image corresponding to the true value image and the geographical location information of the true value image;

[0074] Step 12, obtaining satellite image reference channel image and satellite image geographic location information;

[0075] Step 13, using the Otsu threshold method to segment the satellite image to obtain cloud area and non-cloud area images;

[0076] Step 14, remapping the true value image using Gaussian filtering to obtain a remapped image;

[0077] Step 15, extracting feature points of the cloud area and non-cloud area images and the remapped images;

[0078] Step 16, using SIFT feature point matching algorithm to perform feature matching to obtain matching point pairs;

[0079] Step 17, based on the matching point pairs, use the RANSAC algorithm to remove outlier matching point pairs to obtain accurate matching point pairs;

[0080] Step 18, extracting the location information of the exact matching point pairs according to the true value image location information and the satellite image location information, and converting it into a positioning error;

[0081] Step 19, performing a positioning accuracy test based on the positioning error to obtain a positioning accuracy test result.

[0082] In an embodiment of the present invention, the positioning accuracy test is used to evaluate the accuracy and precision of satellite images. The positioning error, that is, the difference between the satellite image and the true value image, can be obtained through the positioning accuracy test to evaluate the accuracy of the positioning algorithm; the positioning accuracy test result can reflect the accuracy of the satellite image positioning, thereby evaluating the availability and applicability of the satellite image; if the positioning accuracy test result is large, there is a large difference between the position of the satellite image and the position of the true value image, and the positioning algorithm needs to be further improved and optimized; the positioning accuracy test is to evaluate the advantages and disadvantages of the positioning algorithm and provide a basis for improving the positioning algorithm; by analyzing and comparing the positioning error, the problems and shortcomings of the positioning algorithm can be found, and then targeted improvements and optimizations can be made to improve the positioning accuracy of the satellite image; the results of the positioning accuracy test can also be used to compare different positioning algorithms, find the optimal positioning algorithm, and provide more accurate and reliable satellite image positioning.

[0083] like Figure 1 As shown, step 11, obtaining the control point image corresponding to the true value image and the geographical location information of the true value image from the high-precision control point image library, includes:

[0084] Step 111, obtain the true value image from the high-precision control point image library:

[0085] Step 112, searching the corresponding control point images of the true value image according to the high-precision control point image library to obtain the corresponding control point images of the value image;

[0086] Step 113: extracting geographic location information from the true value image and the corresponding control point image to obtain the true value image geographic location information.

[0087] In an embodiment of the present invention, the purpose of obtaining the control point image corresponding to the true value image and the geographical location information of the true value image from the high-precision control point image library is to provide a reference and basis for the subsequent positioning accuracy test; by obtaining the control point image corresponding to the true value image, a reference image of the true value image in the high-precision control point image library can be obtained, and the reference image can be used as a standard for comparison of the true value image; by comparing the true value image and the corresponding control point image, the positioning accuracy and precision of the satellite image can be evaluated; at the same time, the geographical location information is extracted from the true value image and the corresponding control point image, and the geographical location coordinates of the true value image can be obtained; the geographical location coordinates can be used in the subsequent positioning accuracy test, compared with the geographical location of the satellite image, and the positioning error and positioning accuracy are calculated; by comparing the geographical location information of the true value image and the satellite image, the positioning accuracy and positioning error of the satellite image can be evaluated; therefore, the effect of obtaining the control point image corresponding to the true value image and the geographical location information of the true value image from the high-precision control point image library is to provide accurate true value images and geographical location information for the subsequent positioning accuracy test, and then evaluate the positioning accuracy and accuracy of the satellite image.

[0088] like Figure 1 As shown, step 12, obtaining satellite image reference channel images and satellite image geographic location information from the load database to be evaluated, includes:

[0089] Step 121, obtaining a satellite image from a database of loads to be evaluated;

[0090] Step 122, searching the reference channel image of the satellite image according to the database of loads to be evaluated to obtain the reference channel image of the satellite image;

[0091] Step 123, extracting geographic location information from the satellite image reference channel image to obtain satellite image geographic location information.

[0092] In an embodiment of the present invention, a satellite image reference channel image and satellite image geographic location information are obtained from a database of loads to be evaluated to provide a reference and basis for subsequent positioning accuracy inspection; by obtaining a satellite image reference channel image, a reference image of the satellite image in the database of loads to be evaluated can be obtained, and the reference image can be used as a benchmark for comparison of the satellite image; by comparing the satellite image and the reference channel image, the positioning accuracy and precision of the satellite image can be evaluated; at the same time, geographic location information is extracted from the satellite image reference channel image to obtain the geographic location coordinates of the satellite image; the geographic location coordinates can be used in subsequent positioning accuracy inspections, compared with the geographic location of the true value image, and the positioning error and positioning accuracy are calculated; by comparing the geographic location information of the satellite image and the true value image, the positioning accuracy and positioning error of the satellite image can be evaluated.

[0093] like Figure 1 As shown, step 13, using the Otsu threshold method to segment the satellite image to obtain cloud area and non-cloud area images, including:

[0094] Step 131, obtaining a satellite image reference channel image;

[0095] Step 132, converting the satellite image reference channel image into a grayscale image;

[0096] Step 133, calculating a histogram of the grayscale level of each pixel in the grayscale image to obtain the number of pixels at each grayscale level in the image;

[0097] Step 134, dividing the number of pixels at each gray level by the total number of pixels to obtain a probability distribution of pixels at each gray level;

[0098] Step 135, using the Otsu threshold method, setting a threshold to segment the grayscale image into two categories, and calculating the variance between the two categories after segmentation to obtain the between-class variance.

[0099] Between-class variance = w0×w1×(μ0-μ1) 2

[0100] Among them, w0 and w1 are the sum of the pixel probabilities of category 0 and category 1, respectively, and μ0 and μ1 are the average grayscale values ​​of category 0 and category 1, respectively;

[0101] Step 136, according to the inter-class variance, select a threshold that maximizes the inter-class variance as the optimal threshold;

[0102] Step 137, binarize the grayscale image according to the optimal threshold, set the pixel grayscale level greater than the threshold as the cloud area, and set the pixel grayscale level less than or equal to the threshold as the non-cloud area.

[0103] In an embodiment of the present invention, the satellite image is segmented by using the Otsu threshold method, that is, the image is segmented into cloud area and non-cloud area, the information of the cloud area is extracted, and the basis for subsequent cloud detection and positioning accuracy test is provided; the Otsu threshold method is an adaptive threshold selection method, which determines the optimal threshold by calculating the inter-class variance; the optimal threshold can be automatically selected according to the grayscale distribution of the image, so that the variance of the two classes after segmentation is maximized, thereby achieving effective image segmentation; after converting the satellite image reference channel image into a grayscale image, the number of pixels at each grayscale level can be obtained by calculating the grayscale level histogram of each pixel in the grayscale image; then, according to the number of pixels and the total image The number of pixels is calculated, and the probability distribution of pixels at each gray level is calculated; the inter-class variance under different threshold conditions is calculated by traversing the gray level threshold using the Otsu threshold method; the threshold that maximizes the inter-class variance is selected as the optimal threshold; according to the optimal threshold, the gray image is binarized, and the pixels with gray levels greater than the threshold are set as cloud areas, and the pixels with gray levels less than or equal to the threshold are set as non-cloud areas; the cloud area and non-cloud area in the satellite image can be effectively separated by image segmentation using the Otsu threshold method; the obtained cloud area images and non-cloud area images can be used for subsequent cloud detection and positioning accuracy testing, and further analyze and evaluate the cloud conditions and positioning accuracy of satellite images.

[0104] like Figure 1 As shown, step 14, using Gaussian filtering to remap the true value image to obtain a remapped image, includes:

[0105] Step 141, obtaining the control point image corresponding to the true value image;

[0106] Step 142, according to the control point image corresponding to the true value image, read the longitude and latitude range of the true value image, and find the corresponding area in the longitude and latitude coordinate map of the corresponding load;

[0107] Step 143, in the satellite payload corresponding area, find the corresponding coordinate position on the true value image pixel by pixel;

[0108] Step 144, according to each corresponding coordinate on the true value image, use Gaussian filtering to extract the pixel mean, and project the pixel mean to the coordinate of the corresponding position of the satellite payload to obtain a remapped image.

[0109] In an embodiment of the present invention, Gaussian filtering is used to remap the true value image to project the pixel values ​​of the true value image from the position on the true value image to the position on the corresponding satellite payload image to obtain the remapped image. In this way, the pixel values ​​corresponding to the true value image can be obtained on the satellite payload image for subsequent feature extraction and matching; before remapping the true value image, it is first necessary to obtain the control point image corresponding to the true value image; by reading the latitude and longitude range of the true value image and finding the corresponding area in the corresponding payload latitude and longitude coordinate map, the area in the satellite payload image corresponding to the true value image can be determined; then, the corresponding coordinate position on the true value image is found pixel by pixel in the corresponding area of ​​the satellite payload; for each corresponding coordinate on the true value image, a Gaussian filter is used to extract the mean of the surrounding pixels, and the mean is projected to the coordinates of the corresponding position of the satellite payload to obtain the remapped image; the satellite payload image corresponding to the true value image can be obtained, so that the two are consistent in pixel values; through the remapped image, the pixel values ​​corresponding to the true value image can be obtained on the satellite payload image, thereby realizing the comparison and matching between the two; it is very important for the subsequent feature extraction and matching algorithms, which can improve the accuracy and precision of matching and further evaluate the positioning accuracy of satellite images.

[0110] like Figure 1 As shown, step 15 is characterized in that extracting feature points of the cloud area and non-cloud area images and the remapped images includes:

[0111] Step 151, obtaining images of the cloud area and the non-cloud area, and the remapped images;

[0112] Step 152, based on the cloud area and non-cloud area images and the remapped images, detect interest points in the images that are scale- and rotation-invariant by using a Gaussian difference function at different scales to obtain extreme points;

[0113] Step 153, according to the extreme points, determine the position and scale of the key points by fitting a refined model;

[0114] Step 154, assigning a main direction to each key point based on the local gradient direction of the image;

[0115] Step 155, in the neighborhood around the key point, the local gradient of the image is measured at the selected scale, and the descriptor of the key point is generated by calculating the local gradient of the image to obtain the feature points of the cloud area and non-cloud area images and the remapped image.

[0116] In an embodiment of the present invention, feature points of cloud area and non-cloud area images and remapped images are extracted to provide the positions and descriptors of key points for subsequent feature matching, which are used for cloud detection and positioning accuracy inspection. First, cloud area and non-cloud area images and remapped images are obtained. The images contain information about cloud areas and non-cloud areas, as well as images obtained by aligning the true value image with the satellite image through remapping. Then, by using Gaussian difference functions at different scales to detect points of interest in the image, extreme points can be obtained. The extreme points have scale and rotation invariance and can be used as key points. The positions and scales of key points are determined by fitting a sophisticated model, which is achieved by calculating the pixel gradients and Hessian matrices of the areas around the key points. To accurately determine the position and scale of the key points; to make the subsequent descriptor generation process invariant to rotation changes, a main direction is assigned to each key point based on the gradient direction histogram of the area around the key point; finally, the local gradient of the image in the neighborhood around the key point is measured at the selected scale, and the descriptor of the key point is generated by calculating the local gradient of the image; the descriptor contains the feature information of the area around the key point and is used to represent the features of the key point; by extracting the feature points of the cloud area and non-cloud area images and the remapped images, these feature points can be used for matching and comparison in subsequent feature matching; it is helpful for cloud detection and positioning accuracy inspection, and provides the position and descriptor of the key points for evaluating the cloud conditions and positioning accuracy of satellite images.

[0117] like Figure 1 As shown, step 16 is characterized in that feature matching is performed using a SIFT feature point matching algorithm to obtain matching point pairs, including:

[0118] Step 161, obtaining the positions, scales, main directions and descriptors of feature points of the cloud area and non-cloud area images and the remapped images;

[0119] Step 162, calculating the similarity between each feature point of the cloud region and non-cloud region images and all feature points of the remapped image;

[0120] Step 163 , screening feature points according to the similarity threshold to obtain matching point pairs.

[0121] In the embodiment of the present invention, the feature matching using SIFT feature point matching algorithm is to match the feature points of the cloud area and non-cloud area images with the feature points of the remapped image, find the corresponding matching point pairs, and use them for cloud detection and positioning accuracy inspection; first, obtain the position, scale, main direction and descriptor of the feature points of the cloud area and non-cloud area images and the remapped image; the feature points contain the local feature information of the image; then, calculate the similarity of each feature point of the cloud area and non-cloud area images with all the feature points of the remapped image; use Euclidean distance to calculate the similarity; by calculating the similarity between the two feature descriptors, it is possible to find the similarity between each feature point in the cloud area and non-cloud area images. The best matching point in the remapped image; feature points are screened according to the similarity threshold to obtain matching point pairs; by setting the similarity threshold, matching points with lower similarity can be screened out and only matching points with higher similarity can be retained; the accuracy and reliability of matching can be improved; feature matching is performed through the SIFT feature point matching algorithm, and the feature points of the cloud area and non-cloud area images can be matched with the feature points of the remapped image to find the corresponding matching point pairs; matching point pairs can be used for cloud detection and positioning accuracy inspection to evaluate the cloud situation and positioning accuracy of satellite images; the number and quality of matching point pairs can be used as important indicators for evaluating positioning algorithms and image quality.

[0122] like Figure 1 As shown, step 17, based on the matching point pairs, uses the RANSAC algorithm to remove outlier matching point pairs to obtain accurate matching point pairs, including:

[0123] Step 171, obtaining matching point pairs and setting them as data sets;

[0124] Step 172, randomly selecting a minimum required number of data points from the data set as an initial model;

[0125] Step 173, estimating the parameters of the model using the selected data points according to the initial model to obtain an estimated model;

[0126] Step 174, calculating the fitting errors between the remaining matching point pairs in the data set and the estimated model;

[0127] Step 175, according to a preset threshold, marking the points whose fitting errors with the estimated model are less than the threshold as inliers, and marking the points whose fitting errors are greater than the threshold as outliers;

[0128] Step 176, comparing the number of inliers of each estimated model, and selecting the model with the largest number of inliers as the best model;

[0129] Step 177, repeat the above steps until the preset maximum number of iterations is reached or the number of inliers meets the preset conditions, and select the model with the largest number of inliers as the final accurate matching model;

[0130] Step 178, using the final exact matching model, removes outliers and obtains exact matching point pairs.

[0131] In the embodiment of the present invention, the RANSAC algorithm is used to remove outlier matching point pairs, and outliers are removed from matching point pairs in an iterative manner to obtain accurate matching point pairs. , which can improve the accuracy and reliability of matching and reduce the interference of outliers on matching results; first, according to the matching point pairs, it is set as a data set; in the RANSAC algorithm, the data set contains all matching point pairs; then, the minimum required number of data points, such as two data points, are randomly selected from the data set as the initial model; the data points are used to estimate the initial parameters of the model; according to the initial model, the selected data points are used to estimate the parameters of the model to obtain the estimated model; for example, in the case of fitting a straight line, the slope and intercept of the straight line can be estimated by the selected data points; the fitting error between the remaining matching point pairs in the data set and the estimated model is calculated; a measurement method of fitting error can be used, such as the distance from the point to the straight line; according to a preset threshold, the matching point pairs that are fitted to the estimated model are Points with fitting errors less than the threshold are marked as inliers, and points with fitting errors greater than the threshold are marked as outliers; then, the number of inliers of each estimated model is compared, and the model with the largest number of inliers is selected as the best model. The model with a large number of inliers has a higher probability of being the correct model; repeat the above steps until the preset maximum number of iterations is reached or the number of inliers meets the preset conditions. In each iteration, the model with the largest number of inliers is selected as the current best model; finally, the best model is used to remove outliers to obtain accurate matching point pairs; matching point pairs are the accurate matching results after outliers are removed by the RANSAC algorithm; by using the RANSAC algorithm to remove outlier matching point pairs, the interference of erroneous matches and outliers can be removed to obtain more accurate and reliable matching point pairs.

[0132] like Figure 1 As shown, step 18, extracting the location information of the exact matching point pair according to the true value image location information and the satellite image location information, and converting it into a positioning error, includes:

[0133] Step 181, obtaining accurate matching points between the true value image and the satellite image, and obtaining the latitude and longitude position information of the matching points;

[0134] Step 182, according to the latitude and longitude position information of the matching point, find four grid points in the neighborhood of the matching point, and obtain the latitude and longitude coordinates of the four grid points;

[0135] Step 183, calculating the position of the matching point in sub-pixel coordinates according to the latitude and longitude position information of the matching point and the latitude and longitude grid of the payload image;

[0136] Step 184, fitting the elevation information of the matching point in the sub-pixel coordinates according to the sub-pixel coordinates of the matching point;

[0137] Step 185, calculating the longitude and latitude coordinates of the matching point in the sub-pixel coordinates according to the sub-pixel coordinates of the matching point;

[0138] Step 186, subtract the grid coordinate position of the matching point from the position in the true value image to obtain the positioning error in the cross-track and along-track directions.

[0139] In an embodiment of the present invention, based on the geographic location information of the true value image and the geographic location information of the satellite image, the location information of the exact matching point pair is extracted and converted into a positioning error to evaluate the positioning accuracy of the satellite image and determine the positioning difference between the satellite image and the true value image; first, the exact matching points of the true value image and the satellite image are obtained, and the longitude and latitude position information of the matching points is obtained; the matching points are the exact matching results obtained through the previous processing steps; then, based on the longitude and latitude position information of the matching points, four grid points in the neighborhood of the matching point are found, and the longitude and latitude coordinates of the four grid points are obtained; the grid points can be used as reference points to calculate the position of the matching points in sub-pixel coordinates; then, based on the longitude and latitude position information of the matching points and the longitude and latitude grid of the payload image, the sub-pixel coordinates of the matching points can be calculated. Position; the sub-pixel coordinates of the matching points can be obtained by interpolation or other methods; according to the sub-pixel coordinates of the matching points, the elevation information of the matching points is fitted in the sub-pixel coordinates, and the fitting model is used to estimate the elevation value of the matching points; finally, according to the sub-pixel coordinates of the matching points, the longitude and latitude coordinates of the matching points in the sub-pixel coordinates are calculated; the sub-pixel coordinates can be converted into longitude and latitude coordinates by reverse projection or other methods; by subtracting the grid coordinate position of the matching point from the position in the true value image, the positioning error of the matching point in the cross-track and along-track directions can be obtained; the positioning error can be used to evaluate the positioning accuracy of the satellite image and determine the positioning difference between the satellite image and the true value image; the size and direction of the positioning error can provide information about the positioning of the satellite image, which is used to evaluate and improve the accuracy and reliability of the positioning algorithm.

[0140] Specifically, let the latitude and longitude of the matching point be (lat0, log0), and the four grid points (x i ,y i )i=1,2,3,4 correspond to the longitude and latitude coordinates respectively (lat i , log i )i=1,2,3,4, let the sub-pixel coordinates of the grid point be the pair (x0,y0), then

[0141]

[0142] Among them, latx =lat2-lat1,lat y =lat3-lat1,log x =log2-log1,log y =log3-log1

[0143] Therefore, the grid coordinates corresponding to (lat0, log0) are (x0+x1, y0+y1). After obtaining the grid coordinates of the corresponding point pair, their coordinate positions are used for subtraction to obtain the cross-track error.

[0144] like Figure 1 As shown, step 19, performing a positioning accuracy test according to the positioning error to obtain a positioning accuracy test result, includes:

[0145] Step 191, obtaining the positioning errors in the cross-track and along-track directions, and calculating the statistics of the positioning errors;

[0146] Step 192, obtaining a positioning accuracy test result based on the statistics of the positioning error.

[0147] In an embodiment of the present invention, the positioning accuracy of the satellite image can be evaluated by performing a positioning accuracy test based on the positioning error, and a positioning accuracy test result can be obtained, which is used to evaluate the positioning accuracy and reliability of the satellite image; first, the positioning errors in the cross-track and along-track directions are obtained; by subtracting the grid coordinate position of the matching point from the position in the true value image, the positioning errors in the cross-track and along-track directions can be obtained; then, the statistics of the positioning error are calculated; the statistics such as the mean value, standard deviation, maximum value, minimum value of the positioning error can be calculated, and other related statistical indicators can be calculated as needed; the statistics can provide the overall situation and distribution characteristics of the satellite image positioning error; based on the statistics of the positioning error, the positioning accuracy test result can be obtained; by comparing By comparing the statistics of the positioning error with the predetermined accuracy requirements or standards, it is possible to evaluate whether the positioning accuracy of the satellite image meets the requirements; if the positioning error is small and the various statistical indicators meet the accuracy requirements, it can be considered that the positioning accuracy of the satellite image is high; conversely, if the positioning error is large, or some statistical indicators do not meet the requirements, it is necessary to further improve and optimize the positioning algorithm to improve the positioning accuracy of the satellite image; the positioning accuracy test results can provide a quantitative evaluation of the positioning accuracy of the satellite image, help users understand the positioning accuracy of the satellite image, and provide a reliable reference for the application of satellite images; the positioning accuracy test results can be used to determine whether the satellite image is suitable for a specific application scenario and whether the positioning algorithm needs to be further improved and optimized.

[0148] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for analyzing the geographic positioning accuracy of medium-resolution satellite images, characterized in that: The method comprises: Obtain the control point image corresponding to the true value image and the geographical location information of the true value image; Obtain satellite image reference channel images and satellite image geographic location information; The satellite image is segmented using the Otsu threshold method to obtain images of cloud area and non-cloud area; Remap the true value image using Gaussian filtering to obtain a remapped image; Extract feature points of cloud area and non-cloud area images and remapped images; Use SIFT feature point matching algorithm to perform feature matching to obtain matching point pairs; According to the matching point pairs, the RANSAC algorithm is used to remove outlier matching point pairs to obtain accurate matching point pairs; According to the geographic location information of the true value image and the satellite image, the location information of the exact matching point pairs is extracted and converted into the positioning error; The positioning accuracy is tested based on the positioning error to obtain the positioning accuracy test result.

2. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 1, characterized in that: Obtain the control point image corresponding to the true value image and the geographic location information of the true value image from the high-precision control point image library, including: Get the ground truth image from the high-precision control point image library: According to the high-precision control point image library, the corresponding control point image of the true value image is retrieved to obtain the corresponding control point image of the value image; The geographic location information is extracted from the true value image and the corresponding control point image to obtain the true value image geographic location information.

3. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 2, characterized in that: Obtain satellite image reference channel images and satellite image geographic location information from the load database to be evaluated, including: Obtain satellite images from the database of loads to be assessed; According to the load database to be evaluated, the reference channel image of the satellite image is retrieved to obtain the reference channel image of the satellite image; The geographic location information is extracted from the satellite image reference channel image to obtain the satellite image geographic location information.

4. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 3, characterized in that: The satellite image is segmented using the Otsu threshold method to obtain cloud area and non-cloud area images, including: Obtain satellite image benchmark channel images; Convert satellite image base channel images into grayscale images; Calculate the histogram of the gray level of each pixel in the grayscale image to obtain the number of pixels at each gray level in the image; The number of pixels at each gray level is divided by the total number of pixels to obtain the pixel probability distribution at each gray level; Using the Otsu threshold method, set the threshold to segment the grayscale image into two categories, and calculate the variance between the two categories after segmentation to obtain the between-class variance. Between-class variance = w0×w1×(μ0-μ1) 2 Among them, w0 and w1 are the sum of the pixel probabilities of category 0 and category 1, respectively, and μ0 and μ1 are the average grayscale values ​​of category 0 and category 1, respectively; According to the inter-class variance, the threshold that maximizes the inter-class variance is selected as the optimal threshold; According to the optimal threshold, the grayscale image is binarized, and the pixel grayscale level greater than the threshold is set as the cloud area, and the pixel grayscale level less than or equal to the threshold is set as the non-cloud area.

5. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 4, characterized in that: Use Gaussian filtering to remap the true value image to obtain the remapped image, including: Get the control point image corresponding to the true value image; According to the control point image corresponding to the true value image, read the longitude and latitude range of the true value image, and find the corresponding area in the longitude and latitude coordinate map of the corresponding load; In the satellite payload corresponding area, find the corresponding coordinate position on the true value image pixel by pixel; According to each corresponding coordinate on the true value image, Gaussian filtering is used to extract the pixel mean, and the pixel mean is projected to the coordinate of the corresponding position of the satellite payload to obtain the remapped image.

6. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 5, characterized in that: Extract feature points of cloud area and non-cloud area images and remapped images, including: Obtain images of cloud area and non-cloud area, and remapped images; Based on the cloud area and non-cloud area images and the remapped images, the Gaussian difference function is used at different scales to detect the interest points in the image that are invariant to scale and rotation, so as to obtain the extreme points; Based on the extreme points, the location and scale of the key points are determined by fitting a refined model; Based on the local gradient direction of the image, a main direction is assigned to each key point; In the neighborhood around the key point, the local gradient of the image is measured at the selected scale, and the descriptor of the key point is generated by calculating the local gradient of the image to obtain the feature points of the cloud area and non-cloud area images and the remapped image.

7. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 6, characterized in that: Use the SIFT feature point matching algorithm to perform feature matching to obtain matching point pairs, including: Obtain the location, scale, main direction and descriptor of feature points of cloud area and non-cloud area images and remapped images; Calculate the similarity between each feature point of the cloud area and non-cloud area images and all feature points of the remapped image; According to the similarity threshold, feature points are screened to obtain matching point pairs.

8. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 7, characterized in that: According to the matching point pairs, the RANSAC algorithm is used to remove outlier matching point pairs to obtain accurate matching point pairs, including: Get matching point pairs and set them as data sets; Randomly select the minimum required number of data points from the dataset as the initial model; Based on the initial model, the parameters of the model are estimated using the selected data points to obtain an estimated model; Calculate the fitting error between the remaining matching point pairs in the data set and the estimated model; According to the preset threshold, the points whose fitting error with the estimated model is less than the threshold are marked as inliers, and the points whose fitting error is greater than the threshold are marked as outliers; Compare the number of inliers for each estimated model and select the model with the largest number of inliers as the best model; Repeat the above steps until the preset maximum number of iterations is reached or the number of inliers meets the preset conditions, and select the model with the largest number of inliers as the final exact matching model; Use the final exact matching model to remove outliers and obtain accurate matching point pairs.

9. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 8, characterized in that: According to the geographic location information of the true value image and the satellite image, the location information of the exact matching point pairs is extracted and converted into positioning errors, including: Obtain the precise matching points between the true image and the satellite image, and obtain the latitude and longitude position information of the matching points; According to the latitude and longitude position information of the matching point, find four grid points in the neighborhood of the matching point and obtain the latitude and longitude coordinates of the four grid points; According to the latitude and longitude position information of the matching point and the latitude and longitude grid of the payload image, the position of the matching point in sub-pixel coordinates is calculated; According to the sub-pixel coordinates of the matching points, the elevation information of the matching points in the sub-pixel coordinates is fitted; According to the sub-pixel coordinates of the matching points, the longitude and latitude coordinates of the matching points in the sub-pixel coordinates are calculated; The grid coordinate position of the matching point is subtracted from the position in the true value image to obtain the positioning error in the cross-track and along-track directions.

10. The method for analyzing the geographic positioning accuracy of medium-resolution satellite images according to claim 9, characterized in that: Perform positioning accuracy inspection based on positioning error to obtain positioning accuracy inspection results, including: Obtain the positioning errors in the cross-track and along-track directions, and calculate the statistics of the positioning errors; According to the statistics of positioning error, the positioning accuracy test result is obtained.

Citation Information

Patent Citations

  • Natural resource remote sensing mapping image positioning method and system

    CN117994678A

  • Apparatus and method for performing visual localization effectively

    KR102616028B1