A large-resolution difference non-homologous synthetic aperture radar sea ice drift detection method

Abstract

The application provides a kind of large resolution difference non-homologous synthetic aperture radar sea ice drift detection method, comprising: obtaining training sample, constructing super-resolution reconstruction network, and using super-resolution reconstruction network uniform resolution;Based on the high-resolution image pair obtained by super-resolution reconstruction network, feature extraction and coarse matching are carried out;Based on the features after feature extraction and coarse matching, remove the error matching points;Calculate the sea ice drift vector field, obtain complete non-homologous SAR image sea water drift information.The application can realize the full-automatic detection of sea ice drift, and use the advantages of non-homologous SAR high spatial resolution and high time resolution, obtain high-resolution, quasi-real-time sea ice drift field, meet the demand of realizing high-resolution sea ice drift automatic detection from massive SAR data.

Description

Technical Field

[0001] This invention relates to the field of sea ice drift technology in non-homogeneous SAR imagery, and more particularly to a method for detecting sea ice drift using non-homogeneous synthetic aperture radar with large resolution differences. Background Technology

[0002] High-precision sea ice drift tracking relies on near real-time, short-time-lag, and information-complete high-resolution image sequences. However, sea ice image sequences constructed based on the same sensor are limited by the generally low frequency of satellite overflight observations, thus restricting the accuracy of sea ice drift tracking. With the advancement of satellite remote sensing technology, there are currently over 200 Earth observation satellites in orbit worldwide. The continuous development in the quantity and quality of observation satellites, along with the significant improvement in the frequency of overflight and spatial resolution of the same region, provides a data foundation for high-precision sea ice drift tracking research by constructing short-time interval image sequences using non-homogeneous satellite data. Furthermore, deeply exploring the correlation patterns between the characteristics of existing non-homogeneous sea ice data is of great significance. Therefore, near real-time, all-weather monitoring of sea ice drift is of great importance. Thus, there is an urgent need to develop a method for near real-time, all-weather monitoring of sea ice drift.

[0003] Monitoring sea ice drift based on satellite-borne remote sensing image sequences is not a new topic. Over the past three decades, many valuable research results have been achieved internationally in this field. Currently, sea ice drift monitoring research utilizes co-source SAR data. However, existing research shows that near-real-time, all-weather monitoring of sea ice drift is not widely conducted internationally, primarily due to unfavorable observation conditions. Monitoring sea ice drift in polar regions using polar orbit and sun-synchronous orbit satellites is difficult to achieve due to low satellite pass frequencies or dense cloud cover, making it challenging to obtain near-real-time, short-latency, and complete image sequences. Currently, commonly used data for sea ice drift monitoring is co-source data, i.e., image sequences taken by the same type of satellite sensor. This approach has limitations: the revisit time for the same satellite is long, requiring at least a day, resulting in low temporal resolution. Within longer time intervals, velocities caused by short-term events (such as storms) are lost. Using multiple satellites for observation makes it easier to obtain short-interval image sequences of the same region (non-co-source data), significantly improving the accuracy of sea ice drift monitoring.

[0004] Non-homogeneous SAR data exhibit resolution differences, even within the same data source but across different product modes. For example, Radarsat-2 provides 100m resolution images in wide-swath scan mode (SCWA) but only 8m resolution in fine mode (F6F). When the difference is small, the sea ice features in the same area appear similarly; however, significant differences often lead to considerable inconsistencies in details such as sea ice outlines. Sea ice in the same area will exhibit different characteristics in SAR at different resolutions. After performing equal-resolution transformation on aligned-synchronous TerraSAR-X (TSX) and ASAR sea ice data, significant differences in sea ice features were found. These differences mainly stem from details such as sea ice cracks, inter-ice channels, ice toes, and ice ridges. When significant resolution differences exist, directly downsampling a high-resolution image to a low-resolution image results in the loss of some effective features from the high-resolution image. Conversely, upsampling a low-resolution image to a high-resolution image results in numerous false corner features due to distortion. Based on the above analysis, the problems and shortcomings of existing technologies are as follows:

[0005] (1) Current SAR data has the characteristics of wide amplitude and high resolution, and the amount of data has increased significantly. The time resolution of sea ice drift monitoring algorithms needs to be improved.

[0006] (2) At the methodological level, facing a large number of SAR images with resolution differences, it is also necessary to overcome the impact of resolution differences on features, as well as the accuracy of matching the same points at multiple times. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method for detecting sea ice drift using non-homogeneous synthetic aperture radar (SAR) with large resolution differences. This invention enables fully automated detection of sea ice drift and leverages the high spatial and temporal resolution advantages of non-homogeneous SAR to obtain a high-resolution, near-real-time sea ice drift field, meeting the requirement for high-resolution automated sea ice drift detection from massive SAR data.

[0008] The technical means employed in this invention are as follows:

[0009] A method for detecting sea ice drift using non-homogeneous synthetic aperture radar with large resolution differences includes:

[0010] Obtain training samples, construct a super-resolution reconstruction network, and use the super-resolution reconstruction network to unify the resolution;

[0011] Feature extraction and coarse matching are performed on high-resolution image pairs obtained by super-resolution reconstruction networks;

[0012] Based on the features extracted and coarsely matched, incorrect matching points are removed.

[0013] Calculate the sea ice drift vector field to obtain complete seawater drift information from non-originating SAR images.

[0014] Furthermore, the steps of acquiring training samples, constructing a super-resolution reconstruction network, and using the super-resolution reconstruction network to unify the resolution include:

[0015] A super-resolution reconstruction network is constructed, which includes the network input part, the cyclic residual network, and the residual frequency combination.

[0016] A smooth super-resolution residual network based on super-resolution edge guidance is constructed, as shown in the following formula:

[0017]

[0018] Where H and D represent the fuzzy and downsampling models, respectively, and x L The output image of the filter is represented by λ, which is a smoothing factor. The smoothing factor λ is determined by calculating the smoothing coefficient of the input image noise to obtain an image with better resolution. The formula is:

[0019]

[0020] in, The intensity of the selected image patch is given by , N is the number of image patches, μ is the average intensity of the image patches, and n is the fitting parameter.

[0021] The super-resolution image restoration process is based on a smooth super-resolution residual network guided by super-resolution edges:

[0022]

[0023] Furthermore, in the super-resolution reconstruction network:

[0024] The network input part uses an exponentially weighted average edge detector to extract the edge map of the input low-resolution image, and at the same time uses an enhanced LEE filter to filter the low-resolution image to obtain a filtered map, which is input into the network together with the low-resolution image.

[0025] The cyclic residual network progressively restores low-resolution images, filtered images, and edge feature images to high-resolution image signals in different frequency subbands.

[0026] The residual frequency combination combines the estimated residual with the sub-band signal to restore the low-resolution image to a high-resolution image.

[0027] Furthermore, the feature extraction and coarse matching based on the constructed super-resolution reconstruction network include:

[0028] The AKAZE algorithm is used for matching, and the L2 norm is used to measure feature similarity to obtain a coarse match.

[0029] Furthermore, the matching using the AKAZE algorithm, employing the L2 norm to measure feature similarity and obtain a coarse match, specifically includes:

[0030] Reference feature point p r Two feature points, s1 and s2, are determined from all feature points to be registered, where feature point s1 is related to the reference feature point p. r The Euclidean distance error d1 between feature point s2 and reference feature point p is minimized. r The Euclidean distance error between them is the next most significant factor. If the ratio ρ1 of d1 and d2 is less than a given threshold of 0.9, then the reference feature point p is considered to be faulty. r Match with the feature points to be registered.

[0031] Furthermore, the removal of erroneous matching points based on the features extracted and coarsely matched includes:

[0032] The obtained matching points The matching points are compared with those of surrounding points, and if the threshold is exceeded, the matching points are removed.

[0033] Furthermore, the obtained matching points The points are compared with matching points in the surrounding area. If the threshold is exceeded, the matching points are removed. Specifically, this includes:

[0034] When the local velocity of sea ice differs significantly from that of the surrounding area, this point is determined to be a mismatch and can be removed.

[0035] If the ratio is less than a given threshold, the reference feature point is considered to match the feature point to be registered.

[0036] When the velocity of a point exceeds a threshold, it is considered an incorrect match point. The threshold is calculated based on the average distance to surrounding points, and the calculation formula is as follows:

[0037]

[0038] Where, d i d represents the distance between feature matching points around the center matching point c in two matched SAR images. c This represents the distance between the center matching point c in two matched SAR images.

[0039] Furthermore, the calculation of the sea ice drift vector field to obtain complete sea ice drift information from non-originating SAR images includes:

[0040] Let the motion distance be the motion distance between the midpoints of the two matched SAR images;

[0041] The speed of sea ice drift is calculated by dividing the distance traveled by the time interval.

[0042] Compared with the prior art, the present invention has the following advantages:

[0043] 1. The non-synthetic aperture radar sea ice drift detection method with large resolution difference provided by the present invention breaks the long time interval of homogeneous SAR images and realizes near real-time sea ice monitoring by coordinating non-synthetic SAR images.

[0044] 2. The non-homogeneous synthetic aperture radar sea ice drift detection method with large resolution differences provided by the present invention adds a smoothing factor to the super-resolution network based on residual network, which reduces the distortion problem of low resolution images in the process of obtaining high resolution images.

[0045] 3. The non-homogeneous synthetic aperture radar sea ice drift detection method with large resolution differences provided by this invention uses a super-resolution reconstruction algorithm to unify the resolution and obtain high-resolution SAR images for sea ice drift detection, which can obtain more features and improve the accuracy of sea ice drift detection.

[0046] Based on the above reasons, this invention can be widely applied in fields such as sea ice drift in non-homogeneous SAR images. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a flowchart of the method of the present invention.

[0049] Figure 2 This is a schematic diagram of the non-homogeneous sea ice drift monitoring process of the present invention.

[0050] Figure 3 This is a schematic diagram of the network structure of the super-resolution algorithm of this invention.

[0051] Figure 4 A comparison diagram of sea ice drift vector fields using the algorithm of this invention and a low-resolution algorithm, provided for embodiments of this invention.

[0052] Figure 5 A comparison diagram of the distance and direction between the estimated sea ice drift vector and manually retrieved reference data provided in this embodiment of the invention.

[0053] Figure 4In the middle: (a) the algorithm of this invention, (b) the low-resolution algorithm;

[0054] Figure 5 In the text: (a) the algorithm of this invention, (b) the traditional algorithm, (c) the algorithm of this invention, (d) the traditional algorithm; Detailed Implementation

[0055] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0058] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following figures denote similar items; therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0059] In the description of this invention, it should be understood that the orientation or positional relationship indicated by directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" is generally based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this invention and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this invention. The directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0060] For ease of description, spatial relative terms such as "above," "over," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation besides the orientation of the device as described in the figures. For example, if the device in the figures is inverted, a device described as "above" or "above" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein will be interpreted accordingly.

[0061] Furthermore, it should be noted that the use of terms such as "first" and "second" to define components is merely for the purpose of distinguishing the corresponding components. Unless otherwise stated, the above terms have no special meaning and therefore should not be construed as limiting the scope of protection of this invention.

[0062] like Figure 1 , 2 As shown, this invention provides a method for detecting sea ice drift using non-synthetic aperture radar with large resolution differences, comprising:

[0063] S1. Obtain training samples, construct a super-resolution reconstruction network, and use the super-resolution reconstruction network to unify the resolution;

[0064] S2. Based on the high-resolution image pairs obtained by the super-resolution reconstruction network, feature extraction and coarse matching are performed;

[0065] S3. Based on the features extracted and coarsely matched, remove incorrect matching points;

[0066] S4. Calculate the sea ice drift vector field to obtain complete seawater drift information from non-homogeneous SAR images.

[0067] In a preferred embodiment of the present invention, step S1, which involves acquiring training samples, constructing a super-resolution reconstruction network, and using the super-resolution reconstruction network to unify the resolution, includes:

[0068] S11. Obtain training samples. Combine the low-resolution images obtained by downsampling the high-resolution images with the high-resolution image data to form multiple sets of data pairs to form a training set. Then, enhance the data by using rotation data augmentation.

[0069] S12. Construct a super-resolution reconstruction network, which includes a network input part, a cyclic residual network, and a residual frequency combination. In this embodiment, the network input part uses an exponentially weighted average edge detector to extract the edge map of the input low-resolution image, and simultaneously uses an enhanced LEE filter to filter the low-resolution image to obtain a filtered image, which is input into the network along with the low-resolution image. The cyclic residual network gradually restores the low-resolution image, the filtered image, and the edge feature image to high-resolution image signals of different frequency sub-bands. The residual frequency combination combines the estimated residual with the sub-band signal to restore the low-resolution image to a high-resolution image.

[0070] S13. Construct a smooth super-resolution residual network based on super-resolution edge guidance, as shown in the following formula:

[0071]

[0072] Where H and D represent the fuzzy and downsampling models, respectively, and x L For the filter output image, λ is a smoothing factor. Different SAR images have different noise levels; therefore, the smoothing factor λ cannot be a constant. To overcome this problem, this invention proposes an adaptive smoothing factor. The smoothing factor λ is determined by calculating the smoothing coefficient of the input image noise to obtain an image with better resolution. The formula is:

[0073]

[0074] in, The intensity of the selected image patch is given by denoted ...

[0075] S14. Based on a smooth super-resolution residual network guided by super-resolution edges, the super-resolution image restoration process is as follows:

[0076]

[0077] In this embodiment, the super-resolution reconstruction network uses MSE as the loss function:

[0078]

[0079] The flowchart of the super-resolution method is as follows: Figure 3 As shown, the learning rate was set to 0.00001. The proposed model has 20 layers. The number of channels in each convolutional layer is fixed at 64. The filter size is set to 5×5, and the padding size is 1. Finally, the sub-band signals obtained by the recursive residual network are combined to obtain the final high-resolution image. After training, the low-resolution SAR image from the experiment is used as input and passed through the above super-resolution network to generate the corresponding high-resolution image.

[0080] In a specific implementation, as a preferred embodiment of the present invention, step S2, based on the constructed super-resolution reconstruction network, involves feature extraction and coarse matching, including:

[0081] The AKAZE algorithm is used for matching, and the L2 norm is used to measure feature similarity to obtain a coarse match. Specifically, this includes:

[0082] Reference feature point p r Two feature points, s1 and s2, are determined from all feature points to be registered, where feature point s1 is related to the reference feature point p. r The Euclidean distance error d1 between feature point s2 and reference feature point p is minimized. r The Euclidean distance error between them is the next most significant factor. If the ratio ρ1 of d1 and d2 is less than a given threshold of 0.9, then the reference feature point p is considered to be faulty. r Match with the feature points to be registered.

[0083] In a preferred embodiment of this invention, in step S3, based on the features extracted and coarsely matched, incorrect matching points are removed. Sea ice drift is influenced by ocean currents and atmospheric circulation. Simultaneously, objects on the sea surface experience mutual pressure. Therefore, matching points of sea surface objects have a high degree of similarity to surrounding sea surface objects. After coarse matching of SAR image pairs, a method for removing incorrect matches is proposed, taking into account the characteristics of sea ice movement, including:

[0084] The obtained matching points The points are compared with matching points in the surrounding area; if the number exceeds a threshold, the matching point is removed. Specifically, this includes:

[0085] When the local velocity of sea ice differs significantly from that of the surrounding area, this point is determined to be a mismatch and can be removed.

[0086] If the ratio is less than a given threshold, the reference feature point is considered to match the feature point to be registered.

[0087] When the velocity of a point exceeds a threshold, it is considered an incorrect match point. The threshold is calculated based on the average distance to surrounding points, and the calculation formula is as follows:

[0088]

[0089] Where, d i d represents the distance between feature matching points around the center matching point c in two matched SAR images. c This represents the distance between the center matching point c in two matched SAR images.

[0090] In a specific implementation, as a preferred embodiment of the present invention, step S4, calculating the sea ice drift vector field and obtaining complete non-homogeneous SAR image seawater drift information, includes:

[0091] Let the motion distance be the motion distance between the midpoints of the two matched SAR images;

[0092] The speed of sea ice drift is calculated by dividing the distance traveled by the time interval.

[0093] Example

[0094] To verify the feasibility of the sea ice drift model, the sea ice drift velocity vector field was tested on four sets of data. This embodiment uses low-resolution and super-resolution images as input for comparison. The data magnification factor was set to 4 times, and a matching threshold of 0.9 was used. The matching results are shown in Table 1. It can be seen that the method of this invention has more correct matching points than the method using low-resolution images as input.

[0095] Table 1 Feature matching statistics of four sets of SAR experimental data

[0096]

[0097] Taking the third set of data as an example, such as Figure 4 As shown, low-resolution images contain significantly fewer points than high-resolution images. Therefore, using super-resolution images is beneficial for improving sea ice tracking performance. Compared to low-resolution images, super-resolution images contain more drift details; for example, they can detect minute sea ice displacements that are undetectable in low-resolution images. Furthermore, many mismatches are obtained from low-resolution images.

[0098] Because buoy coordinate information is lacking in the Bohai Sea region, manually retrieved ground-based real sea ice vectors were used as data to verify the algorithm of this invention. The ground-based real sea ice vectors are obtained using QGIS software based on human eye-captured feature recognition. While the ground truth is affected by mouse positioning accuracy, this method is reliable. Figure 5The results show that the accuracy of the algorithm of this invention is higher than that of existing methods. After 60 comparisons with manually extracted vectors, the root mean square error is 68m, and the included angle is 2.47°. Compared with traditional methods that reduce resolution, the algorithm of this invention has a higher root mean square error accuracy of 64m and a higher included angle accuracy of 2.77°.

[0099] Registration of SAR images at different resolutions, containing numerous weakly textured targets (such as sea ice and shoals), faces two main challenges: 1) Due to the varying feature performance of images at different resolutions, unifying the resolution introduces numerous false features or results in the loss of true features; 2) Matching weakly textured SAR sea ice targets often leads to repetition and confusion of local structures around feature points, making feature matching difficult. Low feature differentiation increases false matching. Furthermore, speckle noise further complicates the matching process. This invention's algorithm achieves coordination between low-resolution and high-resolution image data, improving the temporal accuracy of sea ice tracking. Experiments show that the sea ice drift vector obtained by this invention effectively approximates the actual sea ice drift results, significantly outperforming the results of the classic manual matching algorithm (MCC) and the drift results obtained from low-resolution data.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting sea ice drift using non-homogeneous synthetic aperture radar with large resolution differences, characterized in that, include: Obtain training samples, construct a super-resolution reconstruction network, and use the super-resolution reconstruction network to unify the resolution, including: To obtain training samples, multiple sets of data pairs are formed by combining low-resolution images obtained from downsampling high-resolution images with high-resolution image data to create a training set, and the data is enhanced by rotation data augmentation. A super-resolution reconstruction network is constructed, which includes the network input part, the cyclic residual network, and the residual frequency combination. A smooth super-resolution residual network based on super-resolution edge guidance is constructed, as shown in the following formula: in, , These represent fuzzy and downsampling models, respectively. The filter outputs the image. It is a smoothing factor, a smoothing factor It is determined by calculating the smoothing coefficient of the noise in the input image to obtain an image with better resolution. The formula is: in, The intensity of the selected image patch, N The number of image patches. The average intensity of the image patch. n These are the fitting parameters; The super-resolution image restoration process is based on a smooth super-resolution residual network guided by super-resolution edges: ； In the super-resolution reconstruction network: The network input part uses an exponentially weighted average edge detector to extract the edge map of the input low-resolution image, and at the same time uses an enhanced LEE filter to filter the low-resolution image to obtain a filtered map, which is input into the network together with the low-resolution image. The cyclic residual network progressively restores low-resolution images, filtered images, and edge feature images to high-resolution image signals in different frequency subbands. The residual frequency combination combines the estimated residual with the sub-band signal to restore the low-resolution image to a high-resolution image. Feature extraction and coarse matching are performed on high-resolution image pairs obtained by super-resolution reconstruction networks; Based on the features extracted and coarsely matched, incorrect matching points are removed. Calculate the sea ice drift vector field to obtain complete seawater drift information from non-originating SAR images.

2. The method for detecting sea ice drift using non-synthetic aperture radar with large resolution differences according to claim 1, characterized in that, The constructed super-resolution reconstruction network performs feature extraction and coarse matching, including: The AKAZE algorithm is used for matching, and the L2 norm is used to measure feature similarity to obtain a coarse match.

3. The method for detecting sea ice drift using non-synthetic aperture radar with large resolution differences according to claim 2, characterized in that, The AKAZE algorithm is used for matching, and the L2 norm is used to measure feature similarity to obtain a coarse match. Specifically, this includes: Reference feature points Two feature points were determined from all the feature points to be registered. and Among them, feature points With reference feature point Euclidean distance error between Minimum, feature point With reference feature point The Euclidean distance error between them is the second largest. and ratio If the value is less than the given threshold of 0.9, then the reference feature point is considered to be... Match with the feature points to be registered.

4. The method for detecting sea ice drift using non-synthetic aperture radar with large resolution differences according to claim 3, characterized in that, The process of removing erroneous matching points based on the features extracted and coarsely matched includes: The obtained matching points The matching points are compared with those of surrounding points, and if the threshold is exceeded, the matching points are removed.

5. The method for detecting sea ice drift using non-synthetic aperture radar with large resolution differences according to claim 4, characterized in that, The obtained matching points The points are compared with matching points in the surrounding area. If the threshold is exceeded, the matching points are removed. Specifically, this includes: When the local velocity of sea ice differs significantly from that of the surrounding area, this point is determined to be a mismatch and can be removed. If the ratio If the value is less than a given threshold, the reference feature point is considered to match the feature point to be registered. When the velocity of a point exceeds a threshold, it is considered an incorrect match point. The threshold is calculated based on the average distance to surrounding points, and the calculation formula is as follows: in, This represents the center matching point in two matched SAR images. The distance between surrounding feature matching points This represents the center matching point in two matched SAR images. The distance between them.

6. The method for detecting sea ice drift using non-synthetic aperture radar with large resolution differences according to claim 1, characterized in that, The calculation of the sea ice drift vector field to obtain complete sea ice drift information from non-originating SAR images includes: Let the motion distance be the motion distance between the midpoints of the two matched SAR images; The speed of sea ice drift is calculated by dividing the distance traveled by the time interval.

Images