CN113160071B - Satellite image automatic geometric correction method, system, media and terminal equipment - Google Patents

Abstract

The invention discloses an automatic geometric correction method, device, storage medium and terminal equipment for a domestic polar small satellite image, which are used for automatically and preferentially screening homologous sensor reference images for automatic geometric correction of a target satellite image; the whole scene image and the image are divided into four parts and nine parts for image enhancement processing, and then homonymous point pairs of the target satellite image and the reference image are extracted to obtain a geographic coordinate file for geometric correction; and carrying out correction precision evaluation on the images corrected by different methods of the target satellite so as to screen out an optimal correction scheme, and correcting the images of the target satellite according to the optimal correction scheme. The method can realize automatic optimal screening of the registration reference images of the homologous sensors, increase the selection of homonymous points by highlighting the detailed characteristics of the polar images through an image enhancement method, and screen an optimal geometric correction scheme according to the image correction precision so as to overcome the defect of low geometric positioning precision of polar small satellites in China.

Description

Satellite image automatic geometric correction method, system, media and terminal equipment

Technical field

The invention relates to the technical field of remote sensing image processing, and in particular to an automatic geometric correction method, system, storage medium and terminal equipment for satellite images.

Background technique

Existing image registration for homologous sensors is mainly carried out using feature-based methods. However, in the face of massive data registration, it is difficult to automatically select the reference images and points of the same name for homologous sensor registration. Especially for polar images, due to the high reflectivity of ice and snow and the sparse texture on the ice sheet surface, the scale-invariant feature transformation algorithm extracts fewer feature points on the ice sheet surface, which makes geometric correction of polar images difficult.

Contents of the invention

The technical problem to be solved by the embodiments of the present invention is to provide an automatic geometric correction method, system, storage medium and terminal equipment for satellite images, which can realize automatic selection and screening of homologous sensor registration reference images, and highlight polar images through image enhancement method Detailed features are used to increase the selection of points with the same name, and the optimal geometric correction scheme is selected based on the image correction accuracy to overcome the shortcomings of low geometric positioning accuracy of my country's polar small satellites.

In order to solve the above technical problems, embodiments of the present invention provide an automatic geometric correction method for satellite images, which is applied to polar small satellites. The method includes:

The reference images used for automatic geometric correction of the target satellite image are selected and screened from three aspects: the distance of image acquisition time, the degree of spatial coverage and the number of point pairs with the same name, in order to obtain the optimal reference image for the geometric correction of the target satellite image;

A preset scheme is used to extract the same-named points of the target satellite image and the MODIS image to obtain the geographical coordinate files for geometric correction of each scheme; wherein, the preset scheme includes extracting the same-named points through the whole scene image, and the image is divided into four parts and nine parts. Partially perform image enhancement processing to extract points of the same name;

The correction accuracy of the target satellite image corrected using the preset scheme is evaluated to select the optimal correction scheme, and the target satellite image is corrected according to the optimal correction scheme.

Furthermore, the MODIS images used for automatic geometric correction of the target satellite image were selected and screened from three aspects: the distance of image acquisition time, the degree of spatial coverage and the number of point pairs with the same name, so as to obtain the optimal reference for the geometric correction of the target satellite image. Image, specifically:

Batch obtain the MODIS images of the target satellite images with few clouds on that day, and create the first data index table of the target satellite images and MODIS images;

Preprocess the target satellite image and MODIS image, use consistent Antarctic projection and 250m resolution, and obtain MODIS image consistent with the geographical range of the target satellite image;

Conduct preliminary screening of the MODIS images, select images whose effective data range covers more than 80% of the target images, and create a second data index table of the target satellite images and MODIS images;

The MODIS images are screened twice, and the points with the same names of the target satellite and MODIS image pairs are extracted according to the second data index table through the feature matching method, and the corresponding MODIS images are selected based on the number of point pairs with the same names. If If there are still multiple scene MODIS data with the same number of point pairs with the same name, the MODIS image close to the shooting time of the target satellite image is used as the final registration reference image, and a third data index table for geometric correction is obtained.

Furthermore, a preset scheme is used to extract the same-name points of the target satellite image and the MODIS image to obtain the geographical coordinate files for geometric correction of each scheme, specifically as follows:

Option 1: After extracting point pairs with the same name through the scale-invariant feature transformation algorithm of the target satellite whole scene image and the corresponding MODIS whole scene image, sort the point pairs with the same name according to their Euclidean distance, eliminate the 10% points with the largest Euclidean distance, and output The real geographical coordinate files of the remaining point pairs with the same name are used as the first corrected geographical coordinate file for correcting the target satellite image;

Option 2: Divide the target image into four parts, and extend each part by 40 pixels to crop the corresponding MODIS image. Each image pair is subjected to piecewise linear stretching and enhancement processing, and then extracted through a scale-invariant feature transformation algorithm. For point pairs with the same name, eliminate 30% of the points with the largest Euclidean distance, output the real geographic coordinate files of the remaining points with the same name, merge the coordinate file here with the first corrected geographic coordinate file, and eliminate duplicate point pairs as the target for correction A second corrected geographic coordinate file of the satellite image;

Option 3: Divide the target image rules into nine parts. The method is the same as Option 2. Obtain the real geographical coordinate file of the remaining points with the same name after excluding 30% of the points with the largest Euclidean distance. Combine the coordinate file here with the first corrected geographical coordinate file. After merging with the second corrected geographical coordinate file and eliminating duplicate point pairs, it becomes the third corrected geographical coordinate file for correcting the target satellite image.

Further, the correction accuracy of the target satellite image corrected by different methods is evaluated to select the optimal correction scheme, and the target satellite image is corrected according to the optimal correction scheme, specifically as follows:

The three corrected geographical coordinate files obtained by the three schemes are used for geometric correction of the target image using the quadratic polynomial method, and the target satellite image corrected by different schemes is obtained;

Using the scale-invariant feature transformation algorithm, the corrected three target satellite images are again extracted from the corresponding MODIS images. After extracting the same points, the Euclidean distance between the point pairs is calculated, the correction accuracy after geometric correction of different schemes is evaluated, and the optimal one is selected. Calibration plan;

Use the optimal correction scheme to perform geometric correction of target satellite images.

In order to solve the above technical problems, embodiments of the present invention also provide an automated geometric correction system for satellite images, including:

The screening module is used to select the reference images for automatic geometric correction of the target satellite image from three aspects: the distance of image acquisition time, the degree of spatial coverage, and the number of point pairs with the same name, so as to obtain the optimal reference image for the geometric correction of the target satellite image. ;

The extraction module is used to extract points of the same name from the target satellite image and the MODIS image using a preset scheme to obtain the geographical coordinate files for geometric correction of each scheme; wherein the preset scheme includes extracting points of the same name and images from the whole scene image. The same points are extracted after image enhancement processing in four parts and nine parts;

The correction module is used to evaluate the correction accuracy of the target satellite image corrected using a preset scheme, to select the optimal correction scheme, and to correct the target satellite image according to the optimal correction scheme.

Further, the screening module is specifically used to:

Batch obtain the MODIS images of the target satellite images with few clouds on that day, and create the first data index table of the target satellite images and MODIS images;

Preprocess the target satellite image and MODIS image, use consistent Antarctic projection and 250m resolution, and obtain MODIS image consistent with the geographical range of the target satellite image;

Conduct preliminary screening of the MODIS images, select images whose effective data range covers more than 80% of the target images, and create a second data index table of the target satellite images and MODIS images;

The MODIS images are screened twice, and the points with the same names of the target satellite and MODIS image pairs are extracted according to the second data index table through the feature matching method, and the corresponding MODIS images are selected based on the number of point pairs with the same names. If If there are still multiple scene MODIS data with the same number of point pairs with the same name, the MODIS image close to the shooting time of the target satellite image is used as the final registration reference image, and a third data index table for geometric correction is obtained.

Further, the extraction module is specifically used to:

Option 1: After extracting point pairs with the same name through the scale-invariant feature transformation algorithm of the target satellite whole scene image and the corresponding MODIS whole scene image, sort the point pairs with the same name according to their Euclidean distance, eliminate the 10% points with the largest Euclidean distance, and output The real geographical coordinate files of the remaining point pairs with the same name are used as the first corrected geographical coordinate file for correcting the target satellite image;

Option 2: Divide the target image into four parts, and extend each part by 40 pixels to crop the corresponding MODIS image. Each image pair is subjected to piecewise linear stretching and enhancement processing, and then extracted through a scale-invariant feature transformation algorithm. For point pairs with the same name, eliminate 30% of the points with the largest Euclidean distance, output the real geographic coordinate files of the remaining points with the same name, merge the coordinate file here with the first corrected geographic coordinate file, and eliminate duplicate point pairs as the target for correction A second corrected geographic coordinate file of the satellite image;

Option 3: Divide the target image rules into nine parts. The method is the same as Option 2. Obtain the real geographical coordinate file of the remaining points with the same name after excluding 30% of the points with the largest Euclidean distance. Combine the coordinate file here with the first corrected geographical coordinate file. After merging with the second corrected geographical coordinate file and eliminating duplicate point pairs, it becomes the third corrected geographical coordinate file for correcting the target satellite image.

Further, the correction module is specifically used to:

The three corrected geographical coordinate files obtained by the three schemes are used for geometric correction of the target image using the quadratic polynomial method, and the target satellite image corrected by different schemes is obtained;

Using the scale-invariant feature transformation algorithm, the corrected three target satellite images are again extracted from the corresponding MODIS images. After extracting the same points, the Euclidean distance between the point pairs is calculated, the correction accuracy after geometric correction of different schemes is evaluated, and the optimal one is selected. Calibration plan;

Use the optimal correction scheme to perform geometric correction of target satellite images.

Embodiments of the present invention also provide a computer-readable storage medium. The computer-readable storage medium includes a stored computer program; wherein, when running, the computer program controls the device where the computer-readable storage medium is located to execute the above. Automatic geometric correction method for satellite images.

An embodiment of the present invention also provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. The processor implements when executing the computer program The above-mentioned automated geometric correction method for satellite images.

Compared with the existing technology, embodiments of the present invention provide an automatic geometric correction method for satellite images. First, the automatic geometric correction of the target satellite image is selected based on three aspects: the distance of image acquisition time, the degree of spatial coverage, and the number of point pairs with the same name. The corrected reference MODIS image is used to obtain the optimal reference image for geometric correction of the target satellite image; then, different methods are used to extract the same-named points between the target satellite image and the MODIS image, including extracting the same-named points through the whole scene image, dividing the image into four parts and After nine parts of image enhancement processing, the same points are extracted to obtain the geographical coordinate files for geometric correction of each plan; finally, the correction accuracy of the target satellite image corrected by different methods is evaluated to screen out the optimal correction plan, and Correct the target satellite image according to the optimal correction scheme. Compared with the existing technology, the present invention can realize automatic selection and screening of homologous sensor registration reference images, and use image enhancement method to highlight the detailed features of polar images to increase the selection of homonymous points, and select the optimal geometry based on the image correction accuracy. Correction scheme to overcome the shortcomings of low geometric positioning accuracy of existing polar small satellites.

Description of the drawings

Figure 1 is a flow chart of an automated geometric correction method for satellite images provided by the present invention;

Figure 2 is a data flow diagram of an automated geometric correction method for satellite images provided by the present invention;

Figure 3 is a diagram showing the effects of different image enhancements in an automated geometric correction method for satellite images provided by the present invention;

Figure 4 is a rendering of an automated geometric correction method for satellite images provided by the present invention;

Figure 5 is a structural block diagram of an automated geometric correction method for satellite images provided by the present invention;

Figure 6 is a structural block diagram of a terminal device provided by the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

It should be noted that the step numbers in the text are only for convenience of explanation of specific embodiments and are not used to limit the execution order of the steps. The method provided in this embodiment can be executed by a relevant server, and the following description takes the server as the execution subject as an example.

As shown in Figures 1 to 4, an embodiment of the present invention provides an automated geometric correction method for satellite images, which is applied to polar small satellites. The method includes steps S11 to S14:

Step S11: Select the reference image for automatic geometric correction of the target satellite image from the three aspects of image acquisition time, spatial coverage, and number of point pairs with the same name, to obtain the optimal reference image for geometric correction of the target satellite image.

Specifically, MODIS images with less cloud cover on the target satellite image that day are obtained in batches, and a first data index table of the target satellite image and MODIS image is created; the target satellite image and MODIS image are preprocessed and a consistent projection is used ( Antarctic projection) and resolution (250m), obtain a MODIS image consistent with the geographical range of the target satellite image through image cropping; conduct a preliminary screening of the MODIS image, and select images whose effective data range covers the target satellite image above a certain target value , and create a second data index table of the target satellite image and the MODIS image; perform a secondary screening of the MODIS image, and extract the same-name points of each target satellite and MODIS image pair according to the second data index table through the feature matching method, Sort and select the corresponding MODIS image according to the number of point pairs with the same name. If there are still multiple scene MODIS data with the same number of point pairs with the same name, select the MODIS image close to the target satellite shooting time as the final registration reference image, and obtain A third data index table for geometric correction is generated.

Furthermore, taking my country's small polar satellite "Jingshi-1" (development code: BNU-1) as an example, batch download the MODIS images (US MODIS reflectance data) corresponding to the images of the BNU-1 satellite for the day, and establish a second The first data index table of the user. BNU-1 data defines the Antarctic projection (set the resolution to 250m) (same as MODIS data); MODIS image extracts radiance in a single band and restores geographical coordinates to output GeoTIFF, defines the Antarctic projection (set the resolution to 250m), MODIS image cropping (250m resolution The corresponding MODIS image is cropped with an extension of 40 pixels (i.e. 10km) of the high-rate BNU-1 image). When the cropped MODIS valid data range/corresponding BNU-1 data range ≥ 0.8, create a second data index table of the BNU-1 satellite image and MODIS image. According to the second data index table, the scale-invariant feature transform (SIFT) matching method is used to extract the same-name point pairs of the BNU-1 satellite image and the corresponding MODIS image. Sort and select the corresponding MODIS image according to the number of point pairs with the same name. If there are still multiple scene MODIS data with the same number of point pairs with the same name, the MODIS data close to the BNU-1 shooting time will be used as the final registration reference image to obtain the final Third data index table used for registration.

Step S12: Use a preset scheme to extract the same points of the target satellite image and the MODIS image to obtain the geographical coordinate files for geometric correction of each scheme; wherein, the preset scheme includes extracting the same points through the whole scene image, dividing the image into four Parts and nine parts are subjected to image enhancement processing to extract identical points.

Among them, scheme one: extract the same-name point coordinates of the same-name point pair through the whole-point image method to obtain a corrected geographical coordinate file, specifically: extract the target satellite whole-scenery image and the corresponding cropped MODIS whole-scenery image through the SIFT algorithm After pairs of points with the same name are sorted according to their Euclidean distance, 10% of the points with the largest Euclidean distance are eliminated, and the real geographical coordinate files of the remaining point pairs with the same name are output as the first method for automatic geometric correction of the target satellite image. Correct geographic coordinates file.

Specifically, the BNU-1 whole scene image and the corresponding cropped MODIS whole scene image (according to the third data index table) use the SIFT algorithm to extract point pairs with the same name, sort them according to the Euclidean distance of the point pairs with the same name, and eliminate the ones with the largest Euclidean distance. 10% of the points, output the real geographical coordinate files of the remaining point pairs with the same name as the first corrected geographical coordinate file used to correct the BNU-1 satellite image.

Among them, Plan 2: Divide the target satellite image into four parts (2*2=4), and extend each part by 40 pixels to crop the corresponding MODIS image, and perform piecewise linear stretching and enhancement processing on each image pair. , and then use the SIFT algorithm to extract point pairs with the same name, eliminate 30% of the points with the largest Euclidean distance, output the real geographical coordinate files of the remaining points with the same name, merge the coordinate file here with the first corrected geographical coordinate file, and eliminate duplicate point pairs as A second corrected geographic coordinate file used to correct the BNU-1 satellite image.

Specifically, please refer to Figure 3. The BNU-1 image is regularly cropped into 4 parts (2*2=4), and each of these four parts is extended by 40 pixels to crop the corresponding MODIS data. The four sets of image pairs were subjected to image segmentation linear stretching and enhancement processing, and then the SIFT algorithm was used to propose name point pairs. The comparison of point pairs with the same name extracted by different stretching methods in Figure 3 is: (a) no stretching; (b) linear stretching; (c) piecewise linear stretching; (d) Gaussian stretching; (e) rectangular Graph equalization stretching; (f) Root mean square stretching (Here we compare the results of six stretching methods: no stretching, linear, piecewise linear, Gaussian, histogram equalization, and root mean square. It is found that piecewise linearity is (more points with the same name can be found by stretching), eliminate 30% of the points with the largest Euclidean distance, and output the real geographical coordinate files of the remaining points with the same name. The coordinate file here is merged with the first corrected geographical coordinate file and the duplicate point pairs are eliminated as the second corrected geographical coordinate file for correcting the BNU-1 satellite image. It is understandable that obtaining points with the same name through image enhancement can highlight the characteristics of ice and snow in the polar regions, thereby finding more points with the same name.

Among them, Plan 3: Divide the target satellite image and MODIS image rules into nine parts (3*3=9), extract point pairs with the same name, eliminate 30% of the points with the largest Euclidean distance, and output the real geographical coordinate files of the remaining points with the same name (Method Same as Scheme 2). Merge the coordinate file here with the first corrected geographical coordinate file and the second corrected geographical coordinate file and eliminate duplicate point pairs as the third corrected geographical coordinate file for correcting the target satellite image.

Specifically, the BNU-1 satellite image and MODIS image are evenly divided into 9 parts for piecewise linear enhancement, and then the same-name point pairs are extracted, and the coordinate file is output after eliminating 30% (the method is the same as the second method). The first corrected geographical coordinate file and the second corrected geographical coordinate file are merged and duplicate point pairs are eliminated to become the third corrected geographical coordinate file for correcting BNU-1 satellite images.

Step S13: Evaluate the correction accuracy of the target satellite image corrected using different schemes to select the optimal correction scheme, and correct the target satellite image according to the optimal correction scheme.

Specifically, the three corrected geographical coordinate files obtained by the three schemes were used for geometric correction of the target satellite image using the quadratic polynomial method to obtain the target satellite image corrected by different schemes; the SIFT algorithm was used to convert the corrected BNU-1 After the satellite images (three types) are again extracted from the corresponding MODIS images, the Euclidean distance between the point pairs is calculated, the correction accuracy after geometric correction of different schemes is evaluated, and the optimal correction scheme is selected; the optimal correction scheme is used for Geometric correction of BNU-1 satellite images.

Furthermore, the three corrected geographical coordinate files obtained by the three schemes were used for geometric correction of BNU-1 satellite images using the quadratic polynomial method to obtain corrected images of different schemes; the corrected BNU-1 satellite images were obtained using the SIFT algorithm. After extracting points with the same name from the satellite images (three types) and the corresponding MODIS images, the Euclidean distance between the point pairs was calculated, the correction accuracy after geometric correction of different schemes was evaluated (Table 1), and the optimal correction scheme was selected.

Table 1 Number of points with the same name and correction accuracy evaluation of three schemes for automatic geometric correction of BNU-1 satellite images

(Take the image captured by the BNU-1 satellite at 09:04:51 on December 18, 2019 as an example)

The automatic geometric correction technology suitable for polar small satellite images is applied to the BNU-1 satellite image for geometric correction. The images before and after correction are superimposed with the corresponding MODIS data in the lower layer, and the effects before and after correction are compared (based on the BNU-1 satellite in December 2019). The image taken at 09:04:51 on March 18th is taken as an example), and the results are shown in Figure 4.

In Figure 4, Figure 4a is the BNU-1 level 0 image, and Figure 4b is the image corrected using the automatic geometric correction technology suitable for domestic polar small satellite images. Figures 4c, e, and g are enlargements of the details in Figure 4a, which are the sea-ice junction, ice sheet, and cloudy parts respectively. Figures 4d, f, and h are enlarged views of the corresponding positions of the image after correction using the technology of the present invention. Judging from the results, the automatic geometric correction technology suitable for domestic polar small satellite images has obvious advantages in the following aspects. First of all, this technology can automatically select and select the registration reference images of homologous sensors (it can collect data from dozens of scenes every day). Select the optimal scene) to avoid manual intervention due to the large amount of satellite data and the need to consider image data range, shooting time, data quality, etc. Secondly, this technology enhances the image in parts, highlighting the detailed features of polar images, allowing the SIFT operator to extract more and more uniform point pairs with the same name on polar images. Third, this technology can select correction solutions based on different images to achieve optimal correction accuracy. Fourth, the geometric accuracy of the image corrected by this technology is significantly improved (due to the increase from 6221.77m to 243.48m), and a series of high-precision registration data products can be automatically produced, providing an effective way to improve and promote the application of polar small satellite data. support and guarantee.

An embodiment of the present invention provides an automatic geometric correction method for satellite images. An embodiment of the present invention provides an automatic geometric correction method for satellite images. First, from three aspects: the time distance of image acquisition, the degree of spatial coverage, and the number of point pairs with the same name. Select and screen the reference MODIS images used for automatic geometric correction of the target satellite image to obtain the optimal reference image for geometric correction of the target satellite image; then, use different methods to extract the same points between the target satellite image and the MODIS image, including through the entire scene The same name is extracted from the image, and the image is divided into four parts and nine parts for image enhancement processing and then the same name points are extracted to obtain the geographical coordinate files for geometric correction of each scheme; finally, the correction accuracy of the target satellite image corrected by different methods is evaluated. To screen out the optimal correction scheme, and correct the target satellite image according to the optimal correction scheme. Compared with the existing technology, the present invention can realize automatic selection and screening of homologous sensor registration reference images, and use image enhancement method to highlight the detailed features of polar images to increase the selection of homonymous points, and select the optimal geometry based on the image correction accuracy. The correction scheme is used to overcome the shortcomings of low geometric positioning accuracy of my country's polar small satellites and meet the needs of practical applications.

As shown in Figure 5, it is a structural block diagram of an automatic geometric correction system for satellite images provided by the present invention. The system includes:

The screening module 21 is used to select the reference images for automatic geometric correction of the target satellite image from three aspects: the distance of image acquisition time, the degree of spatial coverage, and the number of point pairs with the same name, so as to obtain the optimal reference for the geometric correction of the target satellite image. image.

Further, the screening module 21 is specifically used to:

Batch obtain the MODIS images of the target satellite images with few clouds on that day, and create the first data index table of the target satellite images and MODIS images;

Preprocess the target satellite image and MODIS image, use consistent Antarctic projection and 250m resolution, and obtain MODIS image consistent with the geographical range of the target satellite image;

Conduct preliminary screening of the MODIS images, select images whose effective data range covers more than 80% of the target images, and create a second data index table of the target satellite images and MODIS images;

The MODIS images are screened twice, and the points with the same names of the target satellite and MODIS image pairs are extracted according to the second data index table through the feature matching method, and the corresponding MODIS images are selected based on the number of point pairs with the same names. If If there are still multiple scene MODIS data with the same number of point pairs with the same name, the MODIS image close to the shooting time of the target satellite image is used as the final registration reference image, and a third data index table for geometric correction is obtained.

The extraction module 22 is used to extract the same points of the target satellite image and the MODIS image using a preset scheme to obtain the geographical coordinate files for geometric correction of each scheme; wherein the preset scheme includes extracting the same points through the whole scene image, The image is divided into four parts and nine parts, and then the same points are extracted after image enhancement processing.

Further, the extraction module 22 is specifically used to:

Option 1: After extracting point pairs with the same name through the scale-invariant feature transformation algorithm of the target satellite whole scene image and the corresponding MODIS whole scene image, sort the point pairs with the same name according to their Euclidean distance, eliminate the 10% points with the largest Euclidean distance, and output The real geographical coordinate files of the remaining point pairs with the same name are used as the first corrected geographical coordinate file for correcting the target satellite image;

Option 2: Divide the target image into four parts, and extend each part by 40 pixels to crop the corresponding MODIS image. Each image pair is subjected to piecewise linear stretching and enhancement processing, and then extracted through a scale-invariant feature transformation algorithm. For point pairs with the same name, eliminate 30% of the points with the largest Euclidean distance, output the real geographic coordinate files of the remaining points with the same name, merge the coordinate file here with the first corrected geographic coordinate file, and eliminate duplicate point pairs as the target for correction A second corrected geographic coordinate file of the satellite image;

Option 3: Divide the target image rules into nine parts. The method is the same as Option 2. Obtain the real geographical coordinate file of the remaining points with the same name after excluding 30% of the points with the largest Euclidean distance. Combine the coordinate file here with the first corrected geographical coordinate file. After merging with the second corrected geographical coordinate file and eliminating duplicate point pairs, it becomes the third corrected geographical coordinate file for correcting the target satellite image.

The correction module 23 is used to evaluate the correction accuracy of the target satellite image corrected using a preset scheme, to select the optimal correction scheme, and to correct the target satellite image according to the optimal correction scheme.

Further, the correction module 23 is specifically used to:

The three corrected geographical coordinate files obtained by the three schemes are used for geometric correction of the target image using the quadratic polynomial method, and the target satellite image corrected by different schemes is obtained;

Using the scale-invariant feature transformation algorithm, the corrected three target satellite images are again extracted from the corresponding MODIS images. After extracting the same points, the Euclidean distance between the point pairs is calculated, the correction accuracy after geometric correction of different schemes is evaluated, and the optimal one is selected. Calibration plan;

Use the optimal correction scheme to perform geometric correction of target satellite images.

An automated geometric correction system for satellite images provided by an embodiment of the present invention first obtains MODIS data in batches on the day the target satellite is operating, and performs data preprocessing on the MODIS data to obtain preprocessed MODIS data; Process the MODIS data and perform preliminary screening and secondary screening in order to extract the same-name point pairs of the target satellite and MODIS data; extract the same-name point coordinates of the same-name point pairs through the whole-point image and image part enhancement method to obtain corrections Geographical coordinate file; perform correction accuracy evaluation on the corrected geographical coordinate file to screen out the optimal correction plan, and correct the satellite image of the target satellite according to the optimal correction plan. Compared with the existing technology, the present invention can reduce the low positioning accuracy of polar small satellites, realize automatic selection and selection of heterogeneous sensor registration reference images, increase the selection of feature-based points with the same name, and meet the needs of practical applications.

Embodiments of the present invention also provide a computer-readable storage medium. The computer-readable storage medium includes a stored computer program; wherein, when running, the computer program controls the device where the computer-readable storage medium is located to execute the above. Satellite image correction method.

An embodiment of the present invention also provides a terminal device. Refer to Figure 6, which is a structural block diagram of a preferred embodiment of a terminal device provided by the present invention. The terminal device includes a processor 10, a memory 20 and a storage medium. The memory 20 is configured as a computer program executed by the processor 10 , and the processor 10 implements the above-mentioned satellite image correction method when executing the computer program.

Preferably, the computer program can be divided into one or more modules/units (such as computer program 1, computer program 2,...), and the one or more modules/units are stored in the in the memory 20 and executed by the processor 10 to complete the present invention. The one or more modules/units may be a series of computer program instruction segments capable of completing specific functions. The instruction segments are used to describe the execution process of the computer program in the terminal device.

The processor 10 may be a central processing unit (CPU), or other general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), Off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., the general-purpose processor can be a microprocessor, or the processor 10 can also It is any conventional processor. The processor 10 is the control center of the terminal device and uses various interfaces and lines to connect various parts of the terminal device.

The memory 20 mainly includes a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required for a function, etc., and the data storage area can store relevant data, etc. In addition, the memory 20 can be a high-speed random access memory, or a non-volatile memory, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card and a flash memory. Card (Flash Card), etc., or the memory 20 may also be other volatile solid-state storage devices.

It should be noted that the above-mentioned terminal equipment may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the structural block diagram in Figure 6 is only an example of a terminal equipment and does not constitute a limitation on the terminal equipment. It may include more than The illustrations show more or fewer components, or combinations of certain components, or different components.

To sum up, the method, system, storage medium and terminal equipment for automatic geometric correction of satellite images provided by the embodiments of the present invention first obtain the MODIS data of the target satellite on the day of operation in batches, and perform data preprocessing on the MODIS data. To obtain pre-processed MODIS data; perform preliminary screening and secondary screening on the pre-processed MODIS data in order to extract the same-name point pairs of the target satellite and MODIS data; extract the said coordinates of the same-named point pairs to obtain a corrected geographical coordinate file; perform a correction accuracy assessment on the corrected geographical coordinate file to screen out the optimal correction plan, and correct the satellite of the target satellite according to the optimal correction plan The image is corrected. Compared with the existing technology, the present invention can reduce the low positioning accuracy of polar small satellites, realize automatic selection and selection of heterogeneous sensor registration reference images, increase the selection of feature-based points with the same name, and meet the needs of practical applications.

The above are only preferred embodiments of the present invention. It should be noted that those of ordinary skill in the art can also make several improvements and modifications without departing from the technical principles of the present invention. These improvements and modifications It should also be regarded as the protection scope of the present invention.

Claims (6)

1. An automated geometric correction method for satellite images, applied to polar small satellites, characterized in that the method includes: The MODIS images used for automatic geometric correction of target satellite images are selected and screened from three aspects: the distance of image acquisition time, the degree of spatial coverage, and the number of point pairs with the same name, in order to obtain the optimal reference image for geometric correction of target satellite images; specifically: Batch obtain the MODIS images of the target satellite images with few clouds on that day, and create the first data index table of the target satellite images and MODIS images; Preprocess the target satellite image and MODIS image, use consistent Antarctic projection and 250m resolution, and obtain MODIS image consistent with the geographical range of the target satellite image; Conduct preliminary screening of the MODIS images, select images whose effective data range covers more than 80% of the target satellite images, and create a second data index table of the target satellite images and MODIS images; The MODIS images are screened twice, and the points with the same names of the target satellite and MODIS image pairs are extracted according to the second data index table through the feature matching method, and the corresponding MODIS images are selected based on the number of point pairs with the same names. If If there are still multiple scene MODIS data with the same number of point pairs with the same name, the MODIS image close to the shooting time of the target satellite image will be used as the final registration reference image, and a third data index table for geometric correction will be obtained; A preset scheme is used to extract the same-named points of the target satellite image and the MODIS image to obtain the geographical coordinate files for geometric correction of each scheme; wherein, the preset scheme includes extracting the same-named points through the whole scene image, and the image is divided into four parts and nine parts. Partially perform image enhancement processing to extract points of the same name; specifically: Option 1: After using the scale-invariant feature transformation algorithm to extract the same-name point pairs from the target satellite whole-scenery image and the corresponding cropped MODIS whole-scenery image according to the third data index table, sort the same-named point pairs according to their Euclidean distance and eliminate them. For the 10% points with the largest Euclidean distance, output the real geographical coordinate files of the remaining point pairs with the same name as the first corrected geographical coordinate file used to correct the target satellite image; Option 2: Divide the target satellite image into four parts, and extend each part by 40 pixels to crop the corresponding MODIS image. Perform piecewise linear stretching and enhancement processing on each image pair, and then use the scale-invariant feature transformation algorithm. Extract point pairs with the same name, eliminate 30% of the points with the largest Euclidean distance, output the real geographic coordinate files of the remaining points with the same name, merge the coordinate file here with the first corrected geographic coordinate file, and eliminate duplicate point pairs as described for correction The second corrected geographical coordinate file of the target satellite image; Option 3: Divide the target satellite image rules into nine parts. The method is the same as Option 2. Obtain the real geographical coordinate file of the remaining points with the same name after excluding 30% of the points with the largest Euclidean distance. Compare the coordinate file here with the first corrected geographical coordinate. The file and the second corrected geographical coordinate file are merged and duplicate point pairs are eliminated as the third corrected geographical coordinate file for correcting the target satellite image; The three corrected geographical coordinate files obtained by the three schemes are used for geometric correction of the target satellite image using the quadratic polynomial method, and the target satellite image corrected by different schemes is obtained; The correction accuracy of the target satellite image corrected using the preset scheme is evaluated to select the optimal correction scheme, and the target satellite image is corrected according to the optimal correction scheme. 2. The automatic geometric correction method of satellite images as claimed in claim 1, characterized in that the correction accuracy of the target satellite image corrected by different methods is evaluated to select the optimal correction plan, and the corrected method is selected according to the optimal correction method. The correction scheme corrects the target satellite image, specifically: Using the scale-invariant feature transformation algorithm, the corrected three target satellite images are again extracted from the corresponding MODIS images. After extracting the same points, the Euclidean distance between the point pairs is calculated, the correction accuracy after geometric correction of different schemes is evaluated, and the optimal one is selected. Calibration plan; Use the optimal correction scheme to perform geometric correction of target satellite images. 3. An automated geometric correction system for satellite images, characterized in that the system includes: The screening module is used to select and screen MODIS images for automatic geometric correction of target satellite images from three aspects: image acquisition time distance, spatial coverage degree and number of point pairs with the same name, so as to obtain the optimal reference image for geometric correction of target satellite images. ;Specifically: Batch obtain the MODIS images of the target satellite images with few clouds on that day, and create the first data index table of the target satellite images and MODIS images; Preprocess the target satellite image and MODIS image, use consistent Antarctic projection and 250m resolution, and obtain MODIS image consistent with the geographical range of the target satellite image; Conduct preliminary screening of the MODIS images, select images whose effective data range covers more than 80% of the target satellite images, and create a second data index table of the target satellite images and MODIS images; The MODIS images are screened twice, and the points with the same names of the target satellite and MODIS image pairs are extracted according to the second data index table through the feature matching method, and the corresponding MODIS images are selected based on the number of point pairs with the same names. If If there are still multiple scene MODIS data with the same number of point pairs with the same name, the MODIS image close to the shooting time of the target satellite image will be used as the final registration reference image, and a third data index table for geometric correction will be obtained; The extraction module is used to extract points of the same name from the target satellite image and the MODIS image using a preset scheme to obtain the geographical coordinate files for geometric correction of each scheme; wherein the preset scheme includes extracting points of the same name and images from the whole scene image. The image is divided into four parts and nine parts to extract the same points after image enhancement processing; specifically: Option 1: After using the scale-invariant feature transformation algorithm to extract the same-name point pairs from the target satellite whole-scenery image and the corresponding cropped MODIS whole-scenery image according to the third data index table, sort the same-named point pairs according to their Euclidean distance and eliminate them. For the 10% points with the largest Euclidean distance, output the real geographical coordinate files of the remaining point pairs with the same name as the first corrected geographical coordinate file used to correct the target satellite image; Option 2: Divide the target satellite image into four parts, and extend each part by 40 pixels to crop the corresponding MODIS image. Perform piecewise linear stretching and enhancement processing on each image pair, and then use the scale-invariant feature transformation algorithm. Extract point pairs with the same name, eliminate 30% of the points with the largest Euclidean distance, output the real geographic coordinate files of the remaining points with the same name, merge the coordinate file here with the first corrected geographic coordinate file, and eliminate duplicate point pairs as described for correction The second corrected geographical coordinate file of the target satellite image; Option 3: Divide the target satellite image rules into nine parts. The method is the same as Option 2. Obtain the real geographical coordinate file of the remaining points with the same name after excluding 30% of the points with the largest Euclidean distance. Compare the coordinate file here with the first corrected geographical coordinate. The file and the second corrected geographical coordinate file are merged and duplicate point pairs are eliminated as the third corrected geographical coordinate file for correcting the target satellite image; The three corrected geographical coordinate files obtained by the three schemes are used for geometric correction of the target satellite image using the quadratic polynomial method, and the target satellite image corrected by different schemes is obtained; The correction module is used to evaluate the correction accuracy of the target satellite image corrected using a preset scheme, to select the optimal correction scheme, and to correct the target satellite image according to the optimal correction scheme. 4. The satellite image automatic geometric correction system according to claim 3, characterized in that the correction module is specifically used to: Using the scale-invariant feature transformation algorithm, the corrected three target satellite images are again extracted from the corresponding MODIS images. After extracting the same points, the Euclidean distance between the point pairs is calculated, the correction accuracy after geometric correction of different schemes is evaluated, and the optimal one is selected. Calibration plan; Use the optimal correction scheme to perform geometric correction of target satellite images. 5. A computer-readable storage medium, characterized in that the computer-readable storage medium includes a stored computer program; wherein the computer program, when running, controls the device where the computer-readable storage medium is located to execute as claimed in the rights The automatic geometric correction method for satellite images according to any one of claims 1 to 2. 6. A terminal device, characterized in that it includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, it implements: The automatic geometric correction method of satellite images according to any one of claims 1 to 2.