InSAR coseismic deformation downsampling method and system based on image saliency

Abstract

This application discloses an InSAR coseismic deformation downsampling method and system based on image saliency, belonging to the field of surface deformation monitoring. The method comprises: extracting average saliency from the InSAR deformation map, binarizing it, and expanding it to generate a mask file; performing size expansion processing on the near-field region and the multi-scale average saliency map to generate a masked multi-scale average saliency map; performing quadtree sampling on the near-field region to generate a first sampling result, and uniformly sampling the far-field region based on a constant window to generate a second sampling result; and splicing the first and second sampling results to generate a target sampling result. The method can preserve the details of the main deformation zone, suppress the interference of far-field or local deformation, obtain reliable coseismic deformation sampling data, and improve the reliability of subsequent studies such as source parameter inversion. It does not require prior fault information or pre-inversion, thus reducing the complexity of the sampling process and enhancing the applicability of the method.

Description

Technical Field

[0001] The present application belongs to the field of surface deformation monitoring, and specifically relates to an InSAR coseismic deformation downsampling method and system based on image saliency. Background Art

[0002] Synthetic Aperture Radar Interferometry (InSAR) is an advanced radar remote sensing Earth observation technology. One of its most important applications is to obtain surface deformation between observations by repeatedly observing the same area. Surface deformation, as a direct indicator of crustal stress release, is key observational data for understanding the evolution of disasters. InSAR technology has been widely used in surface deformation monitoring. Compared with point-based observation methods such as the Global Navigation Satellite System (GNSS), InSAR provides more comprehensive surface deformation information, supporting more detailed and in-depth research on the causes and impacts of disasters. Coseismic deformation refers to surface displacement at the time of an earthquake. The coseismic deformation field obtained by InSAR has the advantages of wide coverage and high spatial resolution. A typical InSAR image contains millions of observation points. A large number of observation points helps to obtain more detailed deformation information, but also imposes a significant computational cost for parameter inversion.

[0003] Before using the InSAR deformation field for fault parameter inversion, data downsampling is performed. Traditional InSAR coseismic deformation downsampling methods include uniform sampling, resolution-based downsampling, and quadtree sampling. However, these sampling methods have the following shortcomings:

[0004] First, uniform sampling is achieved by averaging the deformation points in a window of a given size. This method is relatively simple and rough. Although it can reduce the amount of data, it will also blur the details of the deformation center.

[0005] Second, resolution-based downsampling methods incorporate initial fault geometry as prior information into the sampling process. They determine the near and far fields based on the distance between the deformation point and the fault, thereby reducing the interference of local deformation in the far field. However, the need for prior information limits the applicability of this method.

[0006] Third, the quadtree sampling algorithm uses deformation gradients to partition windows for sampling. This algorithm can significantly reduce the amount of data without using prior information, effectively preserving key deformation features. Compared with resolution-based downsampling methods, quadtree sampling is more widely used in InSAR seismic research. However, the coseismic deformation field of InSAR is often affected by temporal decorrelation, atmospheric delay, and other factors, resulting in localized deformation, which can lead to high deformation gradient areas in areas far from the deformation center. Quadtree sampling methods that use deformation gradients as window partitioning indicators may sample many unnecessary points in these areas, reducing downsampling efficiency and the accuracy of subsequent parameter inversion. Summary of the Invention

[0007] The purpose of the embodiments of the present application is to provide an InSAR co-seismic deformation downsampling method and system based on image saliency, which can solve the problems of high complexity and low applicability of existing sampling technologies.

[0008] In order to solve the above technical problems, this application is implemented as follows:

[0009] In a first aspect, an embodiment of the present application provides an InSAR co-seismic deformation downsampling method based on image saliency, the method comprising: performing average saliency extraction on a preset InSAR deformation map to obtain a multi-scale average saliency map; performing binarization and dilation processing on the multi-scale average saliency map in sequence to generate a mask file, wherein the mask area with the largest area in the mask file is the near-field area, and the other areas in the mask file except the near-field area are the far-field areas; performing size expansion processing on the mask file and the multi-scale average saliency map of the near-field area respectively to generate a masked multi-scale average saliency map; performing quadtree sampling on the near-field area based on the masked multi-scale average saliency map and the size-expanded InSAR deformation map to generate a first sampling result, and performing uniform sampling on the far-field area based on a constant window to generate a second sampling result; and splicing the first sampling result and the second sampling result to generate a target sampling result.

[0010] As an optional implementation of the first aspect of the present application, a process of performing average significance extraction on a preset InSAR deformation map to obtain a multi-scale average significance map includes: obtaining a two-dimensional matrix image of the InSAR co-seismic deformation to be sampled; setting the sizes of a sliding window and a search window, wherein the size of the sliding window is a sequence and the size of the search window is a fixed constant; expanding the width of the normalized InSAR deformation map outward and filling it with null values; calculating the average value of non-null value points in the sliding window and the search window; calculating the absolute value of the difference between the average values, and assigning the absolute value to all pixels in the sliding window as the significance of each pixel in the sliding window; and moving the sliding window and the search window to traverse the image image to generate the significance of the image image in the deformation field.

[0011] Change the size of the sliding window and repeat the above steps to generate multiple saliency maps of different sizes of the deformation field;

[0012] The average saliency of the image is calculated based on the saliency map of each image to generate a multi-scale average saliency map.

[0013] As an optional implementation of the first aspect of the present application, obtaining a first sampling result includes: performing size expansion processing on a preset InSAR deformation map to generate an expanded InSAR deformation map; performing quadtree sampling based on the expanded InSAR deformation map and the masked multi-scale average saliency map, a set minimum window, a maximum window, and a segmentation threshold, and obtaining four sub-windows after dividing the image into four equal parts; calculating the standard deviation of each sub-window respectively, and iteratively quartering each sub-window that satisfies the standard deviation greater than the set threshold until the standard deviation of the sub-window is less than the set threshold or the sub-window size is equal to the minimum window, stopping the segmentation, and generating multiple segmented sub-windows; calculating the longitude, latitude, and deformation average of each sampling point in the segmented sub-window; and generating the first sampling result based on the longitude, latitude, and deformation average.

[0014] As an optional implementation of the first aspect of the present application, the first sampling result is a matrix of N rows and 3 columns, the first column is longitude, the second column is latitude, and the third column is the deformation average, where N represents the number of sampling points.

[0015] As an optional implementation of the first aspect of the present application, the process of uniformly sampling the far-field area based on a constant window to generate a second sampling result includes: setting a constant window based on the maximum window set by quadtree sampling; segmenting the preset InSAR deformation map based on the constant window to obtain multiple segmented far-field areas; calculating the deformation average of each window in each segmented far-field area; and obtaining the second sampling result based on the deformation average.

[0016] In a second aspect, an embodiment of the present application provides an InSAR co-seismic deformation downsampling system based on image saliency, the system comprising:

[0017] The saliency extraction module is used to extract the average saliency of the preset InSAR deformation map and obtain a multi-scale average saliency map;

[0018] The mask file generation module is used to perform binarization and dilation processing on the multi-scale average saliency map in sequence to generate a mask file. The mask area with the largest area in the mask file is the near-field area, and the mask area with the smallest area in the mask file is the far-field area.

[0019] A size expansion module is used to perform size expansion processing on the mask file and the multi-scale average saliency map of the near-field area respectively to generate a masked multi-scale average saliency map;

[0020] A sampling module is used to perform quadtree sampling on the near-field area based on the masked multi-scale average saliency map and the scaled InSAR deformation map to generate a first sampling result, and to perform uniform sampling on the far-field area based on a constant window to generate a second sampling result;

[0021] The target sampling result generating module is used to combine the first sampling result and the second sampling result to generate the target sampling result.

[0022] In a third aspect, an embodiment of the present application provides an electronic device comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the method of the first aspect.

[0023] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the method of the first aspect are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a flowchart of an InSAR coseismic deformation downsampling method based on image saliency provided in the first embodiment of the present application;

[0025] Figure 2 This is a detailed flowchart of an InSAR coseismic deformation downsampling method based on image saliency provided in the first embodiment of the present application;

[0026] Figure 3 1 is a schematic diagram of obtaining an image saliency map using a multi-scale filtering method provided in the first embodiment of the present application;

[0027] Figure 4 The first embodiment of the present application provides a satellite line-of-sight coseismic deformation map and saliency maps at different scales;

[0028] Figure 5 This is an internal structure diagram of an InSAR co-seismic deformation downsampling system based on image saliency provided in the second embodiment of the present application. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0030] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects and are not used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments of this application can be implemented in an order other than those illustrated or described herein. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0031] In the following, in conjunction with the accompanying drawings, a method and system for downsampling InSAR co-seismic deformation based on image saliency provided by the embodiment of the present application will be described in detail through specific embodiments and application scenarios.

[0032] Example 1

[0033] See also Figure 1 , which is a flowchart of an InSAR co-seismic deformation downsampling method based on image saliency provided by the present invention, including steps S1 to S5.

[0034] Step S1: Perform average saliency extraction on the preset InSAR deformation map to obtain a multi-scale average saliency map.

[0035] Specifically, for the preset InSAR (Interferometric Synthetic Aperture Radar,

[0036] The process of extracting the average significance of the deformation map and obtaining the multi-scale average significance map includes: obtaining a two-dimensional matrix image of the InSAR co-seismic deformation to be sampled; setting the size of the sliding window and the search window, the size of the sliding window is a sequence, and the size of the search window is a fixed constant; expanding the width of the normalized InSAR deformation map outward Then fill with null values; calculate the sliding window and search window The average value of the non-null value points and ; Calculate the significance of each pixel in the sliding window based on the average value ; The absolute value Assigned to sliding window All pixels in the image are taken as their saliency values; the sliding window and the search window are moved to traverse the image to generate the saliency of each image in the deformation field ; Based on the saliency of each image Calculate the average saliency of the image , to generate a multi-scale average saliency map.

[0037] The detailed process of obtaining the multi-scale average saliency map in the above steps is as follows:

[0038] S11. Obtain the InSAR co-seismic deformation image to be sampled.

[0039] Specifically, in step S11 , the deformable image is two-dimensional matrix data, and the two-dimensional matrix includes a plurality of deformable images.

[0040] S12. Use a multi-scale filtering method to obtain the saliency of each pixel in the deformation field.

[0041] See also Figure 4 , represents the satellite line-of-sight coseismic deformation map and the significance map at different scales, Figure 4 Demonstrated the use of different The significance of the values ​​and their average values ​​are obtained, and the sliding window is increased Size It will reflect the saliency at a larger scale, but the resolution of the saliency map will gradually decrease, and the search window The size of is positively correlated with the integrity of the salient region. Specifically, step S12 includes:

[0042] S121. Setting a sliding window The size of , search window The size of ,in, Representation sequence , , Representation sequence length.

[0043] S122, calculating sliding window and search window The average value of the non-null value points and Then calculate the significance value, where the mathematical expression of the significance value is:

[0044]

[0045] Sliding window The significance of all points in .

[0046] S123, move the two windows to traverse the entire image and obtain the saliency value of each deformed image ,in, , moving sliding window and search window , until the entire deformation map is traversed to obtain the saliency value .

[0047] S124. Change sliding window Size , repeat steps S121 to S123 to obtain multiple significance values.

[0048] S125. Calculate the mean of the multiple saliencies in step S124 to obtain the mathematical expression of the average saliency of the image:

[0049]

[0050] in, represents the average saliency of the image, Representation sequence length, Represents the saliency value of each deformed image.

[0051] Step S2: After binarization and dilation processing are performed on the multi-scale average saliency map, a mask file is generated. The mask area with the largest area in the mask file is the near-field area. Except for the near-field area, the other areas in the mask file are far-field areas.

[0052] Specifically, the process of binarizing the multi-scale average saliency map in sequence includes: The threshold is generally set to a value that can better separate the main deformation area. The purpose of binarization is to create a mask file; the expansion process is used to fill the data gaps and make the main deformation area form a whole. The mask file created after binarization and expansion can distinguish between near-field, far-field and local deformation. The mask area with the largest area is the near-field area.

[0053] Step S3: performing size expansion processing on the mask file and the multi-scale average saliency map of the near-field area respectively to generate a masked multi-scale average saliency map.

[0054] Specifically, see Figure 2 On the one hand, the multi-scale average saliency map is expanded to obtain the expanded multi-scale average saliency map, whose size is On the other hand, the multi-scale average saliency map is binarized and expanded in turn, the largest block (the mask area with the largest area) is selected as the near-field area, and the entire image is expanded (the expanded size is also ), and is fused with the expanded multi-scale average saliency map to generate a masked multi-scale average saliency map. Figure 3 Schematic diagram showing the multi-scale filtering method for obtaining image saliency maps.

[0055] Step S4: Based on the masked multi-scale average saliency map and the scaled InSAR deformation map, quadtree sampling is performed on the near-field area to generate a first sampling result, and uniform sampling is performed on the far-field area based on a constant window to generate a second sampling result.

[0056] Furthermore, the specific process of performing quadtree sampling on the near-field area based on the masked multi-scale average saliency map to generate the first sampling result in step S4 includes:

[0057] S41, expand the multi-scale average saliency map and the original deformation map based on the original multi-scale average saliency map to a length and width of 2 n Multi-scale average saliency map of pixels for quadrant segmentation.

[0058] S42, using the cropped full image size as the initial window, using the standard deviation of significance in the initial window As a reference indicator for near-field quadtree sampling, the mathematical expression of the reference indicator is:

[0059]

[0060] in, represents the standard deviation, represents the number of midpoints in the window, Indicates the first The significance value of the point, It represents the mean significance of all non-null points in the window.

[0061] S43, when the window When the value is greater than the set threshold, the window is divided into four sub-windows according to their size, and each sub-window has the same size.

[0062] S44, respectively calculate the annotation deviation corresponding to each of the four sub-windows after segmentation , if in the child window If the value in the sub-window is greater than the set threshold, it will be segmented iteratively. If the value is less than the set threshold or the window is less than the set minimum window size, segmentation will be stopped.

[0063] S45 . Using the divided panes, calculate the average deformation value in each pane and use it as the deformation value of the center point of the pane, and generate a first sampling result based on the deformation value.

[0064] For further information, see Figure 2 The process of generating the first sampling result is as follows: first, the preset InSAR deformation map is resized to generate an expanded InSAR deformation map; quadtree sampling is performed based on the expanded InSAR deformation map and the masked multi-scale average saliency map, the set minimum window, the maximum window, and the segmentation threshold, and the image is divided into four equal parts to obtain four sub-windows; the standard deviation of each sub-window is calculated respectively, and each sub-window with a standard deviation greater than the set threshold is iteratively divided into four equal parts until the standard deviation of the sub-window is less than the set threshold or the sub-window size is equal to the minimum window, thereby generating multiple segmented sub-windows; the longitude, latitude, and deformation average value of the sampling point of each segmented sub-window are calculated; and the first sampling result is generated based on the longitude, latitude, and deformation average value.

[0065] Furthermore, step S4 performs uniform sampling of the far field area based on a constant window to generate a second sampling result, and the process is as follows: setting a constant window based on the maximum window set by quadtree sampling; segmenting the preset InSAR deformation map based on the constant window to obtain multiple segmented far field areas; calculating the deformation average of each window in each segmented far field area; obtaining a second sampling result based on the deformation average, wherein the second sampling result is a matrix of N rows and 3 columns, the longitude of the deformation average is set to the first column of the second sampling result, the latitude of the deformation average is set to the second column of the second sampling result, and the deformation average is set to the third column of the second sampling result, wherein N represents the number of sampling points.

[0066] Step S5: concatenate the first sampling result and the second sampling result to generate a target sampling result.

[0067] Specifically, after obtaining the first sampling result and the second sampling result, since both the first sampling result and the second sampling result are in matrix form, the first sampling result and the second sampling result are concatenated to generate a target sampling result in matrix form.

[0068] This method creates a mask file based on saliency. The specific process involves binarizing and dilating the average saliency. Binarization is used to create the mask file, while the dilation algorithm is used to fill in data gaps and consolidate the main deformation areas. The mask file created through binarization and dilation can distinguish between near-field, far-field, and local deformations, with the largest mask area representing the near-field region.

[0069] The present invention also performs quadtree sampling of significance. The specific process is: quadtree sampling of the near field is performed using the average significance map after masking, and uniform sampling of the far field is performed using a constant large window. In the process of quadtree sampling of significance, the standard deviation of the significance in the window is used. As a reference indicator for near-field quadtree sampling. When the value is greater than the set threshold, the window is divided into four small windows, which are iterated in turn until the value in the partition window is Less than the threshold or the window is smaller than the set minimum window size. As with the traditional quadtree algorithm, the image needs to be cropped or expanded to a length and width of pixels (n=0, 1, 2…) for quartering.

[0070] In summary, the beneficial effects of the InSAR coseismic deformation downsampling method based on image saliency provided by the present invention are:

[0071] First, the sampling method proposed in the present invention is simple and effective. It can retain the details of the main deformation zone, suppress the interference of far-field or local deformation, thereby obtaining reliable co-seismic deformation sampling data and improving the reliability of subsequent studies such as source parameter inversion.

[0072] Second, unlike resolution-based and pre-inversion-based downsampling methods, the sampling method proposed in the present invention does not require prior fault information and pre-inversion, which reduces the complexity of the sampling process and enhances the applicability of the method.

[0073] Example 2

[0074] See also Figure 5 , which shows the internal structure of an InSAR co-seismic deformation downsampling system based on image saliency provided by the present invention, comprising:

[0075] The saliency extraction module 100 is used to perform average saliency extraction on the preset InSAR deformation map to obtain a multi-scale average saliency map;

[0076] The mask file generation module 200 is used to generate a mask file after binarizing and dilating the multi-scale average saliency map. The mask area with the largest area in the mask file is the near-field area. Except for the near-field area, the other areas in the mask file are far-field areas.

[0077] A size expansion module 300 is used to perform size expansion processing on the mask file and the multi-scale average saliency map of the near-field area to generate a masked multi-scale average saliency map;

[0078] A sampling module 400 is configured to perform quadtree sampling on the near-field region based on the masked multi-scale average saliency map to generate a first sampling result, and to perform uniform sampling on the far-field region based on a constant window to generate a second sampling result;

[0079] The target sampling result generating module 500 is configured to concatenate the first sampling result and the second sampling result to generate a target sampling result.

[0080] The beneficial effects of the image saliency-based InSAR coseismic deformation downsampling system provided by this invention are its simplicity and effectiveness. It can preserve details in major deformation zones while suppressing interference from far-field or local deformation, thereby obtaining reliable coseismic deformation sampling data and improving the reliability of subsequent studies such as source parameter inversion. Furthermore, unlike resolution-based and pre-inversion-based downsampling systems, the proposed sampling system does not require prior fault information or pre-inversion, reducing the complexity of the sampling process and enhancing the system's applicability.

[0081] In the embodiments of the present application, an InSAR coseismic deformation downsampling system based on image saliency can be a device, or a component, integrated circuit, or chip in a terminal. The device can be a mobile electronic device or a non-mobile electronic device. For example, the mobile electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), while the non-mobile electronic device can be a server, network attached storage (NAS), personal computer (PC), television, ATM, or self-service machine, etc., without specific limitations in the embodiments of the present application.

[0082] In the embodiments of the present application, an InSAR coseismic deformation downsampling system based on image saliency can be a device having an operating system. The operating system can be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiments of the present application.

[0083] The InSAR co-seismic deformation downsampling system based on image saliency provided in the embodiment of the present application can achieve Figures 1 to 4 In the method embodiment, each process of implementing an InSAR co-seismic deformation downsampling system based on image saliency is not described here in detail to avoid repetition.

[0084] Optionally, an embodiment of the present application also provides an electronic device, including a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, each process of the above-mentioned embodiment of the InSAR co-seismic deformation downsampling method based on image saliency is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0085] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, each process of the above-mentioned embodiment of the InSAR co-seismic deformation downsampling method based on image saliency is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0086] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk.

[0087] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be noted that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0088] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a more preferred embodiment. Based on this understanding, the technical solution of this application, or the part that contributes to the existing technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of this application.

[0089] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

Claims

1. An InSAR coseismic deformation downsampling method based on image saliency, characterized by: include: The method comprises the following steps: performing average significance extraction on a preset InSAR deformation map to obtain a multi-scale average significance map, wherein the method comprises the following steps: obtaining a two-dimensional matrix image of the InSAR co-seismic deformation to be sampled; setting the sizes of a sliding window and a search window, wherein the size of the sliding window is a sequence and the size of the search window is a fixed constant; expanding the width of the normalized InSAR deformation map outward and filling it with null values; calculating the average values ​​of the non-null value points of the sliding window and the search window; calculating the absolute value of the difference between the average values, and assigning the absolute value to the points in the sliding window. all pixel points of the image as the saliency of each pixel point in the sliding window; move the sliding window and the search window to traverse the image to generate the saliency of the image in the deformation field; change the size of the sliding window, repeat the above steps to generate multiple saliency maps of different scales of the deformation field; calculate the average saliency of the image based on the multiple saliency maps of different scales of the deformation field to generate a multi-scale average saliency map; after binarizing and dilating the multi-scale average saliency map in sequence, generate a mask file, wherein the mask area with the largest area in the mask file is the near field area, and the other areas in the mask file except the near field area are the far field areas; performing size expansion processing on the mask file of the near-field area and the multi-scale average saliency map respectively to generate a masked multi-scale average saliency map; Performing quadtree sampling on the near-field region based on the masked multi-scale average saliency map and the scaled InSAR deformation map to generate a first sampling result, and performing uniform sampling on the far-field region based on a constant window to generate a second sampling result; The first sampling result and the second sampling result are concatenated to generate a target sampling result.

2. The InSAR coseismic deformation downsampling method based on image saliency according to claim 1, characterized in that: Obtaining the first sampling result includes: Performing size expansion processing on the preset InSAR deformation map to generate an expanded InSAR deformation map; Performing quadtree sampling based on the expanded InSAR deformation map, the masked multi-scale average saliency map, the set minimum window, the maximum window, and the segmentation threshold, and dividing the image into four equal parts to obtain four sub-windows; Calculating the standard deviation of each sub-window respectively, and iteratively dividing each sub-window whose standard deviation is greater than a set threshold into four equal parts until the standard deviation of the sub-window is less than the set threshold or the sub-window size is equal to the minimum window size, thereby generating a plurality of divided sub-windows; Calculating the longitude, latitude and deformation average of each of the segmented sub-window sampling points; A first sampling result is generated based on the longitude, latitude and deformation average value.

3. The InSAR coseismic deformation downsampling method based on image saliency according to claim 2, characterized in that: The first sampling result is a matrix of N rows and 3 columns, the first column is the longitude, the second column is the latitude, and the third column is the deformation average value, where N represents the number of sampling points.

4. The InSAR coseismic deformation downsampling method based on image saliency according to claim 1, characterized in that: The process of uniformly sampling the far-field area based on a constant window to generate a second sampling result includes: Setting a constant window based on the maximum window set by the quadtree sampling; Segmenting a preset InSAR deformation map based on the constant window to obtain a plurality of segmented far-field regions; Calculating the deformation average of each window in each of the segmented far-field regions; A second sampling result is obtained based on the deformation average value.

5. An InSAR coseismic deformation downsampling system based on image saliency, characterized by: include: A saliency extraction module is configured to perform average saliency extraction on a preset InSAR deformation map to obtain a multi-scale average saliency map, wherein the process of performing average saliency extraction on the preset InSAR deformation map to obtain a multi-scale average saliency map comprises: obtaining a two-dimensional matrix image of the InSAR co-seismic deformation to be sampled; setting the sizes of a sliding window and a search window, wherein the size of the sliding window is a sequence and the size of the search window is a fixed constant; expanding the width of the normalized InSAR deformation map outward and filling it with null values; calculating the average values ​​of non-null value points in the sliding window and the search window; calculating the absolute value of the difference between the average values ​​and assigning the absolute value to all pixels in the sliding window as the saliency of each pixel in the sliding window; moving the sliding window and the search window to traverse the image image to generate the saliency of the image image in the deformation field; changing the size of the sliding window and repeating the above steps to generate multiple saliency maps of the deformation field at different scales; and calculating the average saliency of the image image based on the multiple saliency maps of the deformation field at different scales to generate a multi-scale average saliency map. a mask file generating module, configured to perform binarization and dilation processing on the multi-scale average saliency map in sequence to generate a mask file, wherein the mask region with the largest area in the mask file is the near-field region, and the regions other than the near-field region in the mask file are the far-field regions; a size expansion module, configured to perform size expansion processing on the near-field region and the multi-scale average saliency map respectively, so as to generate a masked multi-scale average saliency map; a sampling module, configured to perform quadtree sampling on the near-field region based on the masked multi-scale average saliency map and the scaled InSAR deformation map to generate a first sampling result, and to perform uniform sampling on the far-field region based on a constant window to generate a second sampling result; The target sampling result generating module is used to combine the first sampling result and the second sampling result to generate a target sampling result.

6. An electronic device, characterized in that: The invention comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, the steps of an InSAR coseismic deformation downsampling method based on image saliency are implemented as described in any one of claims 1 to 4.

7. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the InSAR coseismic deformation downsampling method based on image saliency are implemented as described in any one of claims 1 to 4.

Images