A method and device for automatically selecting reference points for interferometric radar unwrapping

Abstract

The present invention relates to an automatic selection method and device for interferometric radar unwrapping reference points, the method comprising: calculating the phase derivative variance of each interferogram, and binarizing the phase derivative variance to obtain a plurality of binarized phase derivative variance maps; setting a first sliding window in each binarized phase derivative variance map; selecting a stable region in a plurality of binarized phase derivative variance maps according to the proportion of stable pixels in each first window and the average coherence of the first pixels; setting a second sliding window in the stable region corresponding to the coherence coefficient map; selecting a target window according to the average coherence of the second pixel, and obtaining the reference pixel and the value of the reference pixel in the target window; when it is judged that the value of the reference pixel meets a preset condition, using the reference pixel as a phase unwrapping reference point; when it is judged that the value of the reference pixel does not meet the preset condition, reselecting a new stable region. The method avoids the selection of unwrapping reference points being affected by human experience errors.

Description

Technical Field

[0001] The invention belongs to the technical field of radar monitoring, and in particular relates to an automatic selection method and a selection device for interferometric radar unwrapping reference points. Background Art

[0002] Synthetic Aperture Radar Interferometry (InSAR) technology combines Synthetic Aperture Radar (SAR) and Interferometric-Survey technology to obtain the tiny deformation of the observed object by using the phase information in the SAR data. With the upgrade of SAR satellites and the development of InSAR technology itself, InSAR technology has been widely used in the research related to the acquisition of surface deformation. The absolute phase difference between two SAR images can be used to estimate the displacement of the radar line of sight (LOS).

[0003] However, in practical applications, due to the existence of phase ambiguity, the phase value directly obtained is a entangled phase, and the phase ambiguity must be removed by phase unwrapping to obtain the true phase. When performing phase unwrapping, a unified spatial reference is required. The specific approach is usually to select a unified spatial reference for all interferograms and remove it from the interferogram stack.

[0004] In recent years, radar satellites have developed rapidly, the satellite revisit cycle has gradually shortened, the number of available SAR images has increased rapidly, and the number of corresponding interferograms has also increased. While this has promoted the development of InSAR technology, it has also caused the problem of "data explosion", making it more difficult to complete efficient InSAR data processing in a short time. At present, there are some software for InSAR analysis, but some data processing links of time-series InSAR technology rely on the operator's experience, and the parameter selection is somewhat blind, which not only makes the data processing process more cumbersome, but also inevitably introduces human errors. Therefore, it is very necessary to optimize the important links of time-series InSAR technology, such as related issues in phase unwrapping, to improve the efficiency and accuracy of its data processing.

[0005] At present, the research on phase unwrapping methods is common, but there are few studies on the selection of phase unwrapping reference points. The current methods for selecting phase unwrapping reference points are mainly to select high coherence points in the interference pattern by manual experience or to specify a fixed reference point (generally the starting point of the image), and use the selected reference point as the spatial reference benchmark for phase unwrapping. However, the manual experience selection method is time-consuming and laborious, and relies more on human experience. It is highly subjective and easily affected by empirical errors. When there is an error in the phase unwrapping reference point, it will also have different degrees of impact on the results of InSAR deformation measurement; although the method of selecting a fixed reference point is simple, the coherence of the reference point will be limited, and the reference point coordinates are located at the edge of the image, which is easy to cause local deformation, resulting in greater uncertainty in the phase unwrapping results. Summary of the invention

[0006] In order to solve the above problems existing in the prior art, the present invention provides a method and device for automatically selecting reference points for interferometric radar unwrapping. The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0007] The embodiment of the present invention provides a method for automatically selecting reference points for interferometric radar unwrapping, comprising the steps of:

[0008] S1, calculating the phase derivative variance of each interference pattern in each pixel neighborhood, and binarizing the phase derivative variance to divide the pixels in the interference pattern into stable pixels and unstable pixels, thereby obtaining a plurality of binarized phase derivative variance maps;

[0009] S2. setting a first sliding window in each of the binary phase derivative variance maps;

[0010] S3, selecting stable regions from the plurality of binary phase derivative variance maps according to the proportion of stable pixels in each first window during the sliding process of the first sliding window and the average coherence of the first pixels;

[0011] S4, setting a second sliding window in the stable area corresponding to the coherence coefficient graph;

[0012] S5, selecting a target window from a plurality of second windows obtained by sliding the second sliding window according to the average coherence of the second pixels, and obtaining a reference pixel in the target window and a value of the reference pixel;

[0013] S6. When it is determined that the value of the reference pixel meets the preset condition, the reference pixel is used as a phase unwrapping reference point; when it is determined that the value of the reference pixel does not meet the preset condition, return to step S3 and reselect a new stable area.

[0014] In one embodiment of the present invention, step S1 comprises:

[0015] S11, calculating the phase derivative variance of each interference pattern in each pixel neighborhood:

[0016]

[0017] Among them, Z m,n is the phase derivative variance, k is the length of the first sliding window centered on the phase center point (m,n), (i,j) is the value of point (i,j) in the k×k window centered on point (m,n), is the winding phase gradient in the x direction, is the winding phase gradient in the y direction, is the mean value of the winding phase gradient in the k×k window in the x direction, is the mean of the winding phase gradient in the k×k window in the y direction;

[0018] S12, excluding edge pixels and water areas in each of the interference patterns to obtain a target surface area of ​​each of the interference patterns;

[0019] S13. Set the pixels in the target surface area whose phase derivative variance is less than or equal to the phase derivative variance threshold as the stable pixels, and set the pixels in the target surface area whose phase derivative variance is greater than the phase derivative variance threshold as the unstable pixels, to obtain the plurality of binary phase derivative variance maps.

[0020] In one embodiment of the present invention, the phase derivative variance threshold comprises 0.1.

[0021] In one embodiment of the present invention, step S2 comprises:

[0022] The first sliding window is set in each of the binary phase derivative variance maps according to the terrain undulation of the region where the interference pattern is located.

[0023] In one embodiment of the present invention, when the terrain undulation is greater than or equal to 150 m, the size of the first sliding window is 8×8-32×32 pixels; when the terrain undulation is less than 150 m, the size of the first sliding window is 32×32-64×64 pixels.

[0024] In one embodiment of the present invention, step S3 includes:

[0025] S31, counting the proportion of the stable pixels in each of the first sliding windows during the sliding process of the first sliding window, and selecting the first window with the largest proportion of the stable pixels as a candidate stable region;

[0026] S32. When the number of the candidate stable regions is 1, use the candidate stable region as the stable region; when the number of the candidate stable regions is greater than 1, select the candidate stable region with the largest average coherence of the first pixel as the stable region.

[0027] In one embodiment of the present invention, step S5 includes:

[0028] S51, calculating the average value of the interference coherence coefficients of all pixels in the second window to obtain the average coherence of the second pixels;

[0029] S52, selecting the second window with the largest average coherence of the second pixels as the target window, and selecting any pixel in the target window as the reference pixel, and taking the average coherence of the second pixels of the target window as the value of the reference pixel.

[0030] In one embodiment of the present invention, step S6 includes:

[0031] When it is determined that the value of the reference pixel is greater than or equal to the reference point threshold, taking the reference pixel as the phase unwrapping reference point;

[0032] When it is determined that the value of the reference pixel is less than the reference point threshold, the process returns to step S3 to reselect a new stable region.

[0033] In one embodiment of the present invention, the reference point threshold comprises 0.8.

[0034] Another embodiment of the present invention provides an automatic selection device for reference points of interferometric radar unwrapping, comprising:

[0035] A binary phase derivative variance map acquisition module is used to calculate the phase derivative variance of each interference pattern in each pixel neighborhood, and binarize the phase derivative variance to divide the pixels in the interference pattern into stable pixels and unstable pixels, thereby obtaining a plurality of binary phase derivative variance maps;

[0036] A first sliding window setting module, used for setting a first sliding window in each of the binarized phase derivative variance maps;

[0037] A stable region selection module, used for selecting stable regions in the plurality of binary phase derivative variance maps according to the proportion of stable pixels in each first window during the sliding process of the first sliding window and the average coherence of the first pixels;

[0038] A second sliding window setting module, used to set a second sliding window in the stable area corresponding to the coherence coefficient map;

[0039] A reference pixel selection module, used for selecting a target window from a plurality of second windows obtained by sliding the second sliding window according to the average coherence of the second pixels, and obtaining the reference pixels in the target window and the values ​​of the reference pixels;

[0040] The judgment module is used to use the reference pixel as a phase unwrapping reference point when it is judged that the value of the reference pixel meets the preset condition; when it is judged that the value of the reference pixel does not meet the preset condition, return to the stable area selection module to reselect a new stable area.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] The present invention calculates the phase derivative variance of each interference pattern and binarizes it, determines the stable area by the proportion of stable pixels, sets a second sliding window in the stable area, and automatically obtains the pixels in the target window as phase unwrapping reference points in combination with the average coherence of the second pixels. Since the method simultaneously considers the interference phase and coherence coefficient information, stable non-isolated high-coherence pixels can be automatically obtained as phase unwrapping reference points, avoiding the influence of empirical errors caused by artificial experience in the selection of unwrapping reference points in phase unwrapping. The result can be used as a spatial reference point in the phase unwrapping process, which greatly improves the accuracy, reliability and precision of the phase unwrapping results, while reducing time and labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of a flow chart of a method for automatically selecting reference points for interferometric radar unwrapping provided by an embodiment of the present invention;

[0044] Figure 2 A schematic diagram of an automatic selection device for interferometric radar unwrapping reference points provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0046] Embodiment 1

[0047] See also Figure 1 , Figure 1 A flow chart of an automatic reference point selection method for interferometric radar unwrapping provided by an embodiment of the present invention. The automatic reference point selection method for interferometric radar unwrapping is used to select spatial reference points in the phase unwrapping process, wherein the phase difference obtained from the interference pattern is actually only the main value, and its value range is between (-π, +π]. To obtain the true phase difference, an integer multiple of 2π must be added or subtracted from this value. This process is called phase unwrapping.

[0048] The interferometric radar unwrapping reference point automatic selection method comprises the steps of:

[0049] S1. Calculate the phase derivative variance of each interference pattern in each pixel neighborhood, and binarize the phase derivative variance to divide the pixels in the interference pattern into stable pixels and unstable pixels, and obtain a number of binarized phase derivative variance maps.

[0050] Specifically, firstly, a number of N SAR images are obtained, and a group of interferograms are obtained by randomly combining the N SAR images in pairs, and the interferograms with the group of interferograms are obtained. Figure 1 A set of coherence maps corresponding to each other is obtained, and the average value of the coherence coefficients of the set of coherence maps is calculated to obtain a coherence coefficient map; at the same time, a set of intensity maps can be obtained from N SAR images, and the intensity maps are correlated with the interference map and coherence map. Figure 1 One-to-one, the corresponding intensity map has the same coordinates as the interference map and coherence map.

[0051] On the basis of the above, step S1 specifically includes:

[0052] S11. Calculate the phase derivative variance of each interference pattern in each pixel neighborhood.

[0053] Specifically, according to the relationship between the residual point and the interference phase, the interference phase in the area where the residual point exists in the interference pattern is discontinuous. The phase derivative variance reflects the change of the surface phase by describing the rate of change of the surface phase. When the value of the phase derivative variance is very small, it indicates that the phase in this area is continuous and relatively stable; if the value of the phase derivative variance is large, it indicates that the phase in this area is discontinuous and changes dramatically. The phase stable area in the interference pattern can be obtained by the phase derivative variance, thereby avoiding the residual point. The phase derivative variance at the wrapped phase center point (m,n) is defined as:

[0054]

[0055] Among them, Z m,n is the phase derivative variance, k is the length of the first sliding window centered on the phase center point (m,n), (i,j) is the value of point (i,j) in the k×k window centered on point (m,n), is the winding phase gradient in the x direction, is the winding phase gradient in the y direction, is the mean value of the winding phase gradient in the k×k window in the x direction, is the mean of the winding phase gradient in the k×k window in the y direction.

[0056] Furthermore, for a number of interference patterns, the above formula can be used to calculate the phase derivative variance of each interference pattern one by one in each pixel neighborhood (k×j window).

[0057] S12. Eliminate edge pixels and water areas in each interference pattern to obtain a target surface area of ​​each interference pattern.

[0058] Specifically, considering that the edge of the image is prone to deformation, which may cause interference fringes on the edge of the interference pattern, the phase derivative variance value of the 10×10 edge pixels in each interference pattern is set to a larger value, which is greater than the set phase derivative variance threshold, such as setting the value to 10, so that the edge pixels are regarded as unstable pixels and excluded.

[0059] At the same time, the pixels with amplitude less than 0.1 in the intensity map are regarded as water areas. In each interference map, the pixel value with the same coordinates as the water area in the intensity map is set to a value greater than the set phase derivative variance threshold, such as 10, so that the water area is regarded as an unstable area and excluded.

[0060] After excluding the edge pixels and water areas of each interference pattern, the target surface area of ​​each interference pattern is obtained.

[0061] S13. Set the pixels in the target surface area whose phase derivative variance is less than or equal to the phase derivative variance threshold as stable pixels, and set the pixels in the target surface area whose phase derivative variance is greater than the phase derivative variance threshold as unstable pixels, to obtain several binary phase derivative variance maps.

[0062] Specifically, the phase derivative variance of the target surface area of ​​each interference pattern is counted, and a phase derivative variance threshold is set, and the pixel value in the target surface area whose phase derivative variance is greater than the phase derivative variance threshold is set to 0 to represent an unstable pixel, and the pixel value less than or equal to the phase derivative variance threshold is set to 1 to represent a stable pixel, thereby obtaining a number of binary phase derivative variance maps, and the number of the several binary phase derivative variance maps is the same as the number of interference patterns.

[0063] In a specific embodiment, the phase derivative variance threshold may be set to 0.1, 0.2, 0.3, etc., preferably 0.1.

[0064] S2. Setting a first sliding window in each binarized phase derivative variance map.

[0065] Specifically, the first sliding window is used to retrieve stable regions. The larger the sliding window, the more stable pixels there are in each window, but the proportion of stable pixels in the window will decrease dramatically. Therefore, it is unreliable to continue selecting reference pixels in a larger window. A window that is too small may cause the obtained stable region to be too fragmented, making it difficult to determine the stable region.

[0066] Therefore, the first sliding window can be set in each binary phase derivative variance map according to the terrain undulation of the area where the interference pattern is located, that is, the size of the first sliding window is first set according to the terrain undulation of the area where the interference pattern is located, and then the first sliding window is set according to the size in each binary phase derivative variance map.

[0067] In a specific embodiment, when the terrain undulation is greater than or equal to 150 m, that is, the area where the interference pattern is located is an area with large terrain undulation, the size of the first sliding window is 8×8-32×32 pixels; when the terrain undulation is less than 150 m, that is, the area where the interference pattern is located is an area with flat terrain, the size of the first sliding window is 32×32-64×64 pixels.

[0068] Optionally, when the region where the interference pattern is located is a region with relatively large terrain fluctuations, the size of the first sliding window is set to 8×8 pixels to obtain a stable region in the interference pattern.

[0069] S3, selecting stable regions from a number of binary phase derivative variance maps according to the proportion of stable pixels in each first window during the sliding process of the first sliding window and the average coherence of the first pixels. Specifically comprising the steps of:

[0070] S31. Count the proportion of stable pixels in each first window, and select the first window with the largest proportion of stable pixels as a candidate stable region.

[0071] Specifically, for each binary phase derivative variance map, a first sliding window is used to slide pixel by pixel to obtain multiple first windows, and then the proportion of stable pixels in each first window is counted, and the first window with the largest proportion of stable pixels is selected as a candidate stable area.

[0072] S32. When the number of candidate stable regions is 1, the candidate stable region is used as the stable region; when the number of candidate stable regions is greater than 1, the candidate stable region with the largest average coherence of the first pixel is selected as the stable region.

[0073] It is understandable that if there is only one first window with the largest proportion of stable pixels, the first window is used as a stable region. If the proportions of stable pixels in multiple first windows are all the maximum values, that is, the proportions of stable pixels in multiple first windows are the same, there are multiple preferred stable regions at this time, and it is necessary to further combine the coherence to preferentially select the first window with the highest average coherence of the first pixel as the stable region. Among them, the average coherence of the first pixel is obtained by calculating the sum of the coherence (coherence coefficient) of all pixels in each candidate stable region and taking the average value.

[0074] S4. Setting a second sliding window in the stable region corresponding to the coherence coefficient graph.

[0075] Specifically, in order to avoid uncertainty caused by isolated high coherence pixel values, a second sliding window is further set in the stable region determined in step S3, and the size of the second sliding window is smaller than the size of the stable region. Optionally, the second sliding window is selected as 1 / 2 of the size of the stable region window.

[0076] S5. Select a target window from a plurality of second windows obtained by sliding the second sliding window according to the average coherence of the second pixels, and obtain reference pixels and values ​​of the reference pixels in the target window.

[0077] Specifically, according to the relationship between residual points and interference coherence, the area with residual points in the interference pattern is often incoherent. Therefore, after determining the stable area in the study area, high coherence points that are less affected by incoherence noise are selected as reference points. Interference coherence indicates the degree of similarity between two SAR images and is an important feature of SAR images. The interference coherence coefficient γ is used to evaluate the quality of the interference phase image. The calculation formula of the interference coherence coefficient is:

[0078]

[0079] Where γ represents the interference coherence coefficient, M is the main image data set, S is the auxiliary image data set, M and S are both zero-mean complex Gaussian random variables, E[·] represents the mathematical expectation, S * represents the conjugate complex number of S.

[0080] In InSAR technology, the value range of γ is [0,1]. When the coherence coefficient value is 0, it means that the two SAR images are completely unrelated; when the coherence coefficient value is 1, it means that the two SAR images are completely related. The coherence coefficient diagram can be used to judge the quality of the interference fringes and the degree of noise disturbance in the interference diagram. The larger the value of the coherence coefficient, the higher the similarity between the two SAR images, and the better the quality of the interference fringes generated by the interference diagram. Therefore, the point with the largest coherence coefficient in the stable region is selected as the reference point.

[0081] Specifically, step S5 includes the steps of:

[0082] S51, calculating the average value of the interference coherence coefficients of all pixels in the second window to obtain the average coherence of the second pixel.

[0083] Specifically, in the stable region, the second sliding window is used to slide pixel by pixel to obtain a plurality of second windows, and then the sum of the interference coherence coefficients of all pixels in each second window is calculated and the average value is taken to obtain the average coherence of the second pixels in each second window.

[0084] S52, selecting the second window with the largest average coherence of the second pixels as the target window, and selecting any pixel in the target window as the reference pixel, and taking the average coherence of the second pixels of the target window as the value of the reference pixel.

[0085] Specifically, the average coherence of the second pixel of the second window is retrieved, and the second window with the largest average coherence of the second pixel is selected as the target window; then, any pixel in the target window is taken as a reference pixel, and the average coherence of the second pixel of the target window, that is, the mean of the coherence coefficients of all pixels, is taken as the value of the reference pixel.

[0086] In a specific embodiment, the upper left pixel, lower left pixel, upper right pixel, and upper right pixel in the target window can be selected as reference pixels, but the selection of reference pixels in this embodiment is not limited to this, and any pixel that can represent the coordinates of the target window can be used as a reference pixel.

[0087] S6. When it is determined that the value of the reference pixel meets the preset conditions, the reference pixel is used as a phase unwrapping reference point; when it is determined that the value of the reference pixel does not meet the preset conditions, return to step S3 and reselect a new stable area.

[0088] Specifically, the value of the selected reference pixel is judged in combination with the overall coherence of the interference pattern. When the value of the reference pixel is judged to be greater than or equal to the reference point threshold, the reference pixel is used as the phase unwrapping reference point; when the value of the reference pixel is judged to be less than the reference point threshold, return to step S3 to reselect a new stable area.

[0089] Specifically, reselecting a new stable region includes: selecting the first window with the least proportion of stable pixels as a new candidate stable region; when the number of new candidate stable regions is 1, selecting the new candidate stable region as a new stable region; when the number of new candidate stable regions is greater than 1, selecting the new candidate stable region with the largest average coherence of the first pixel as the new stable region.

[0090] Specifically, since the value range of the interference coherence coefficient γ is [0,1], the larger the interference coherence coefficient, the brighter the pixel and the higher the coherence. Therefore, the reference point threshold can be set to a more appropriate value according to the actual coherence of the area where the interference pattern is located. For example, the reference point threshold can be set to 0.8.

[0091] This embodiment is based on a set of interference patterns obtained from N SAR images, calculates the phase derivative variance of each pixel for each interference pattern, and binarizes the phase derivative variance by setting a phase derivative variance threshold, and then sets a first sliding window to obtain pixels with stable phase in each first window; then counts the proportion of stable pixels in each first window of all interference patterns and sorts them, sets a second sliding window again in the first window with the largest proportion of stable pixels, and calculates the average coherence in each second window in the coherence coefficient map to ensure that the reference pixel is not an isolated high coherence point, and finally obtains high coherence pixels in the phase stable area as phase unwrapping reference points. Since the algorithm considers the interference phase and coherence coefficient information at the same time, stable high coherence pixels can be automatically obtained as phase unwrapping reference points, avoiding the influence of empirical errors caused by manual experience, and the results can be used as spatial reference points in the phase unwrapping process, which greatly improves the accuracy, reliability and precision of the phase unwrapping results, while reducing time and labor costs.

[0092] Embodiment 2

[0093] Based on Example 1, see Figure 2 , Figure 2 A schematic diagram of an automatic selection device for interferometric radar unwrapping reference points provided by an embodiment of the present invention.

[0094] The interferometric radar unwrapping reference point automatic selection device comprises a binary phase derivative variance map acquisition module, a first sliding window setting module, a stable region selection module, a second sliding window setting module, a reference pixel selection module and a judgment module.

[0095] Specifically, the binary phase derivative variance map acquisition module is used to calculate the phase derivative variance of each interference map in the neighborhood of each pixel, and binarize the phase derivative variance to divide the pixels in the interference map into stable pixels and unstable pixels, so as to obtain a number of binary phase derivative variance maps. The first sliding window setting module is used to set a first sliding window in each binary phase derivative variance map. The stable region selection module is used to select a stable region in a number of binary phase derivative variance maps according to the proportion of stable pixels in each first window during the sliding process of the first sliding window and the average coherence of the first pixels. The second sliding window setting module is used to set a second sliding window in the stable region corresponding to the coherence coefficient map. The reference pixel selection module is used to select a target window from a number of second windows obtained by sliding the second sliding window according to the average coherence of the second pixel, and obtain the reference pixel and the value of the reference pixel in the target window. The judgment module is used to use the reference pixel as a phase unwrapping reference point when the value of the reference pixel meets the preset conditions; when the value of the reference pixel does not meet the preset conditions, return to the stable area selection module to reselect a new stable area.

[0096] The device for automatically selecting reference points for interferometric radar unwrapping provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effects are similar, which will not be repeated here.

[0097] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A method for automatically selecting reference points for interferometric radar unwrapping, characterized in that: Includes steps: S1, calculating the phase derivative variance of each interference pattern in each pixel neighborhood, and binarizing the phase derivative variance to divide the pixels in the interference pattern into stable pixels and unstable pixels, thereby obtaining a plurality of binarized phase derivative variance maps; S2. setting a first sliding window in each of the binary phase derivative variance maps; S3, selecting a stable region in the plurality of binary phase derivative variance maps according to the proportion of stable pixels in each first window during the sliding process of the first sliding window and the average coherence of the first pixels; step S3 comprises: S31, counting the proportion of the stable pixels in each first window during the sliding process of the first sliding window, and selecting the first window with the largest proportion of the stable pixels as a candidate stable region; S32, when the number of the candidate stable regions is 1, selecting the candidate stable region as the stable region; when the number of the candidate stable regions is greater than 1, selecting the candidate stable region with the largest average coherence of the first pixels as the stable region; S4, setting a second sliding window in the stable area corresponding to the coherence coefficient graph; S5, selecting a target window from a plurality of second windows obtained by sliding the second sliding window according to the average coherence of the second pixel, and obtaining the reference pixel in the target window and the value of the reference pixel; Step S5 includes: S51, calculating the average value of the interference coherence coefficients of all pixels in the second window to obtain the average coherence of the second pixel; S52, selecting the second window with the largest average coherence of the second pixel as the target window, and selecting any pixel in the target window as the reference pixel, and taking the average coherence of the second pixel of the target window as the value of the reference pixel; S6. When it is determined that the value of the reference pixel meets the preset conditions, the reference pixel is used as the phase unwrapping reference point; when it is determined that the value of the reference pixel does not meet the preset conditions, return to step S3 to reselect a new stable area; step S6 includes: when it is determined that the value of the reference pixel is greater than or equal to the reference point threshold, the reference pixel is used as the phase unwrapping reference point; when it is determined that the value of the reference pixel is less than the reference point threshold, return to step S3 to reselect a new stable area.

2. The method for automatically selecting reference points for interferometric radar unwrapping according to claim 1, characterized in that: Step S1 includes: S11. Calculate the phase derivative variance of each interference pattern in each pixel neighborhood: Among them, Z m,n is the phase derivative variance, k is the length of the first sliding window centered on the phase center point (m,n), (i,j) is the value of point (i,j) in the k×k window centered on point (m,n), is the winding phase gradient in the x direction, is the winding phase gradient in the y direction, is the mean value of the winding phase gradient in the k×k window in the x direction, is the mean of the winding phase gradient in the k×k window in the y direction; S12, excluding edge pixels and water areas in each of the interference patterns to obtain a target surface area of ​​each of the interference patterns; S13. Set the pixels in the target surface area whose phase derivative variance is less than or equal to the phase derivative variance threshold as the stable pixels, and set the pixels in the target surface area whose phase derivative variance is greater than the phase derivative variance threshold as the unstable pixels, to obtain the plurality of binary phase derivative variance maps.

3. The method for automatically selecting reference points for interferometric radar unwrapping according to claim 2, characterized in that: The phase derivative variance threshold includes 0.

1.

4. The method for automatically selecting reference points for interferometric radar unwrapping according to claim 1, characterized in that: Step S2 includes: The first sliding window is set in each of the binary phase derivative variance maps according to the terrain undulation of the region where the interference pattern is located.

5. The method for automatically selecting reference points for interferometric radar unwrapping according to claim 4, characterized in that: When the terrain relief is greater than or equal to 150 m, the size of the first sliding window is 8×8-32×32 pixels; when the terrain relief is less than 150 m, the size of the first sliding window is 32×32-64×64 pixels.

6. The method for automatically selecting reference points for interferometric radar unwrapping according to claim 1, characterized in that: The reference point threshold comprises 0.

8.

7. An automatic selection device for interferometric radar unwrapping reference points, characterized in that: The method for automatically selecting reference points for interferometric radar unwrapping according to any one of claims 1 to 6 comprises: A binary phase derivative variance map acquisition module is used to calculate the phase derivative variance of each interference pattern in each pixel neighborhood, and binarize the phase derivative variance to divide the pixels in the interference pattern into stable pixels and unstable pixels, thereby obtaining a plurality of binary phase derivative variance maps; A first sliding window setting module, used for setting a first sliding window in each of the binarized phase derivative variance maps; A stable region selection module, used for selecting stable regions in the plurality of binary phase derivative variance maps according to the proportion of stable pixels in each first window during the sliding process of the first sliding window and the average coherence of the first pixels; A second sliding window setting module, used to set a second sliding window in the stable area corresponding to the coherence coefficient map; A reference pixel selection module, used for selecting a target window from a plurality of second windows obtained by sliding the second sliding window according to the average coherence of the second pixels, and obtaining the reference pixels in the target window and the values ​​of the reference pixels; The judgment module is used to use the reference pixel as a phase unwrapping reference point when it is judged that the value of the reference pixel meets the preset condition; when it is judged that the value of the reference pixel does not meet the preset condition, return to the stable area selection module to reselect a new stable area.

Images