An on-board inversion method for ocean internal wave SAR images

Abstract

The invention discloses an on-board inversion method for ocean internal wave SAR images, belonging to the technical field of internal wave automatic detection and internal wave feature inversion, comprising: based on analyzing the characteristics of ocean SAR images, a method combining image power spectrum characteristics with improved elliptical domain normalized Radon transform is proposed to automatically detect ocean areas containing internal waves, and invert the wavelength, wave direction and wave speed parameters of the internal waves in the areas; the method combining image power spectrum characteristics with improved elliptical domain normalized Radon transform can automatically detect ocean internal wave areas in SAR images, and quickly invert the wavelength, wave direction and wave speed of the internal waves, has the advantages of high automation, high detection rate and low false alarm rate, and can provide an important reference for on-board automatic detection and feature inversion of ocean internal waves in SAR images.

Description

Technical Field

[0001] The invention relates to the technical field of automatic internal wave detection and internal wave feature inversion, and in particular to an on-board inversion method for ocean internal wave SAR images. Background Art

[0002] Near-ocean internal waves are waves that occur inside the ocean. In terms of ocean internal wave detection, timely discovery of internal waves and reporting of their position, propagation direction, amplitude, wave speed, etc. are of great significance to avoiding marine disasters caused by internal waves. Therefore, it is urgently necessary to develop internal wave onboard automatic processing technology.

[0003] In terms of internal wave parameter inversion, the existing technology mentions the use of porous wavelet transform and multi-scale fusion to extract ocean internal wave characteristics, the use of KDV equation simulation signal and SAR signal fitting method to invert internal wave parameters, VMD-based inversion of SAR ocean internal wave parameters, and wavelet transform-based extraction of SAR image ocean internal wave characteristics. These solutions cannot automatically identify areas containing internal waves during the processing process, and require manual selection of internal wave image areas for fine processing.

[0004] The existing technology also uses the EMD method to extract ocean internal wave parameters, and improves the EMD method, proposing a new method of using EMD endpoint extension to invert internal wave parameters, and using integrated empirical mode decomposition (EEMD) to overcome mode aliasing and boundary effect problems. These solutions all require manual selection of internal solitary wave image profiles along the internal wave propagation direction, and then further processing and inversion.

[0005] There is also an automatic detection method for ocean internal waves in satellite SAR images based on Faster R-CNN in the prior art, which has a high detection accuracy. However, the detection method based on deep learning network has large computational complexity, high hardware cost, and complex model design, which is not suitable for the processing environment on the satellite with relatively simple hardware logic and relatively weak computing power. There is also a method of estimating internal wave parameters using elliptical domain normalized Radon transform, but this method has a very high false alarm rate when encountering images interfered by other ocean phenomena or islands. Later, a method for automatically extracting internal wave parameters by combining two-dimensional continuous wavelet transform and elliptical domain normalized Radon transform was further proposed, but the two-dimensional continuous wavelet transform is highly dependent on the selection of wavelet basis. When the SAR internal wave morphology is changeable, the accuracy of internal wave detection is reduced, and the two-dimensional continuous wavelet transform has a large computational complexity and slow calculation speed, which is difficult to meet the requirements of fast processing on the satellite.

[0006] Therefore, it is urgent to design a technical design for automatic detection and feature inversion of ocean internal waves in SAR images to meet the technical requirements of fast automatic detection and feature inversion on satellites. Summary of the invention

[0007] In order to solve the above problems, the present application provides an on-board inversion method for ocean internal wave SAR images, comprising the following steps:

[0008] Based on the synthetic aperture radar SAR satellite, after correcting the image incident angle radiation non-uniformity, the sliding window size is set according to the image resolution and the typical spatial distribution range of the internal solitary wave, and each sliding window image block is processed in blocks;

[0009] Perform Fourier transform on the SAR block image, obtain the normalized power spectrum, make a rough judgment, and obtain the block containing internal waves;

[0010] For the blocks containing internal waves, the elliptical domain normalized Radon transform is performed to extract the Radon domain peaks and valleys corresponding to the wave lines of the internal waves, and further judge whether there are internal waves in the image area to eliminate false alarms;

[0011] The blocks containing internal waves are geographically marked and the internal wave characteristic parameters are output, and the onboard detection results are transmitted to the ground.

[0012] Preferably, in the process of performing block processing on each sliding window image block, based on the set sliding window size, sliding is performed in a partially overlapping manner to perform block processing on each sliding window image block.

[0013] Preferably, in the process of making a rough judgment, according to the wavelength range of the internal wave, the power spectrum values ​​within the wavelength range are accumulated as the power spectrum characteristic value, and by setting the power spectrum characteristic threshold, it is roughly judged that the SAR block image contains internal waves.

[0014] Preferably, in the process of extracting the Radon domain peaks and valleys corresponding to the wave lines of the internal waves, the elliptical domain normalized Radon transform is improved to obtain an improved Radon transform image, and the Radon domain peaks and valleys corresponding to the wave lines of the internal waves are extracted, wherein the improvement process of the elliptical domain normalized Radon transform is as follows: after supplementing the missing part of the block containing the internal waves with the mean of the remaining part, the bright spots or dark spots corresponding to the bright lines and dark lines of the image are obtained by the elliptical domain Radon transform, the maximum value area and the minimum value area are obtained according to the local maximum value and local minimum value method, the elliptical domain normalized Radon transform is performed, and the maximum value and the minimum value points are obtained using a unified threshold value, and the Radon transform image is normalized again.

[0015] Preferably, in the process of further judging whether internal waves exist in the image region, further judging whether internal waves exist in the image region is performed based on the number and angle difference of the extracted Radon domain peaks, peak points or valleys, valley points.

[0016] Preferably, in the process of eliminating false alarms, the internal wave stripes are located, and in the improved Radon transform diagram, the center point of the extreme value area close to the horizontal axis is obtained, that is, the Radon domain extreme point corresponding to the stripes in the image is located. If there are more than two extreme value points close to the horizontal axis in the improved Radon domain, it is determined that internal waves exist. If there are no extreme value points that meet the conditions, the area is determined to be a false alarm, and the false alarm is eliminated.

[0017] Preferably, in the process of obtaining the internal wave characteristic parameters of the block containing the internal wave, the internal wave characteristic parameters include the internal wave direction, the internal wave wavelength, and the internal wave amplitude and wave velocity.

[0018] Preferably, in the process of extracting the internal wave direction, the internal wave direction is extracted based on the bending direction of the internal wave line of the Radon domain image that conforms to the distribution of the internal wave line.

[0019] Preferably, in the process of obtaining the internal wave wavelength, the internal wave wavelength is obtained according to the distance between the front and the front point or the valley and the valley point in the Radon domain.

[0020] Preferably, in the process of obtaining the internal wave amplitude and velocity, the current block position is calculated according to the position of SAR image acquisition, and the jump layer characteristics are obtained in combination with the SAR acquisition time. The jump layer characteristics are combined with the internal wave wavelength characteristics to calculate the internal wave amplitude and velocity.

[0021] The present invention discloses an on-board inversion system for ocean internal wave SAR images, which is used to implement the above-mentioned on-board inversion method, comprising:

[0022] The block processing module is used to correct the image incident angle radiation non-uniformity based on the synthetic aperture radar SAR satellite, set the sliding window size according to the image resolution and the typical spatial distribution range of the internal solitary wave, and perform block processing on each sliding window image block;

[0023] The internal wave identification module is used to perform Fourier transform on the SAR block image, obtain the normalized power spectrum, make a rough judgment, and obtain the block containing the internal wave;

[0024] The false alarm elimination module is used to perform elliptical domain normalized Radon transform on the blocks containing internal waves, extract the Radon domain peaks and valleys corresponding to the wave lines of the internal waves, further determine whether there are internal waves in the image area, and eliminate false alarms;

[0025] The detection result output module is used to geographically mark the blocks containing internal waves and output the internal wave characteristic parameters, and transmit the on-board detection results to the ground.

[0026] The present invention discloses the following technical effects:

[0027] The present invention adopts a method of combining power spectrum characteristics with improved elliptical domain normalized Radon transform to quickly detect internal wave areas and invert internal wave characteristic parameters, providing a key algorithm model for SAR onboard processing and providing fast data support for SAR satellite ocean applications.

[0028] The present invention can automatically detect the ocean internal wave area in SAR images and quickly invert the internal wave wavelength, wave direction and wave speed. It has the advantages of high automation, high detection rate and low false alarm rate, and can provide an important reference for the on-board automatic detection and feature inversion of ocean internal waves in SAR images. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0030] Figure 1 It is a schematic diagram of the internal wave on-board detection process of the present invention;

[0031] Figure 2 The internal wave wavelength statistics of the historical satellite SAR internal wave images described in the present invention;

[0032] Figure 3 It is the elliptical domain Radon transform when the data of a part of the image area is missing as described in the present invention, wherein (a) represents the internal wave image with image data loss, and (b) represents the elliptical domain Radon transform when the image data is lost;

[0033] Figure 4 It is the elliptical domain Radon transform when there is interference from other ocean phenomena described in the present invention, wherein (a) represents the internal wave image when there is interference from other ocean phenomena, and (b) represents the elliptical domain Radon transform when there is interference from other ocean phenomena;

[0034] Figure 5 It is the elliptical domain Radon transform after missing area supplementation described in the present invention, wherein (a) represents the missing area supplementation mean processing image, and (b) represents the improved elliptical domain Radon transform;

[0035] Figure 6 are the minimum values ​​of the normalized Radon transform in the elliptical domain of two typical disturbed images described in the present invention, where (a) represents the internal wave image after the missing area is supplemented ( Figure 5 a) Radon domain minimum, (b) represents the disturbed image ( Figure 4 a) Radon domain minimum;

[0036] Figure 7 It is the Radon transform minimum value after removing the large area in the present invention;

[0037] Figure 8 is the corresponding extreme point of the internal wave stripe extracted by the improved Radon transform described in the present invention, (a) represents Figure 3 The internal wave corresponds to the minimum region of the Radon domain, (b) represents Figure 4 Internal waves correspond to the minimum region of the Radon domain;

[0038] Fig. 9 1 is an example of the corresponding relationship between the internal wave propagation direction and the Radon domain of the present invention, wherein (a) indicates that the internal wave line bends inward and the center of gravity of the Radon domain moves outward, and (b) indicates that the internal wave line bends outward and the center of gravity of the Radon domain moves inward;

[0039] Fig.10 The SAR image and internal wave area automatic detection results of the present invention, wherein (a) represents a 50-meter resolution on-board imaging SAR image acquired by a new satellite, and (b) represents the SAR image acquired by the method of the present invention. Fig.10 (a) Internal wave region detected by image;

[0040] Fig.11 The SAR image and internal wave area automatic detection results of the present invention, wherein (a) represents a 50-meter resolution on-board imaging SAR image acquired by a new satellite, and (b) represents the SAR image acquired by the method of the present invention. Fig.11 (a) Internal wave region detected by image. DETAILED DESCRIPTION

[0041] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0042] like Figure 1-11As shown, the present invention aims at the demand for rapid automatic detection and feature inversion of internal waves on board, and proposes a method of combining power spectrum features with improved elliptical domain normalized Radon transform to rapidly detect internal wave areas and invert internal wave feature parameters, which can provide a key algorithm model for SAR onboard processing and provide rapid data support for SAR satellite ocean applications.

[0043] The basic steps of automatic detection and feature inversion of SAR images of ocean internal waves mentioned in the present invention are as follows: Figure 1 Its main processes and mechanisms are:

[0044] (1) Due to the different SAR incident angles along the azimuth, the radiation amount will be large at small incident angles and small at large incident angles, resulting in a situation where one side of the image is bright and the other side is dark. Therefore, in order to avoid the interference of radiation unevenness on subsequent detection, the image incident angle radiation unevenness is first corrected.

[0045] (2) Set the sliding window size according to the image resolution and the typical spatial distribution range of internal solitary waves. In order to avoid missing areas during the sliding process, the sliding can be performed in a partially overlapping manner, and each sliding window image block is subsequently processed in blocks. The internal waves observed in SAR images are generally internal solitary waves, and the spatial distribution range of internal solitary waves is from hundreds of square kilometers to thousands of square kilometers. In order to better locate the internal wave area, a medium-sized window pane can be used, such as a sliding window of about 25 square kilometers.

[0046] (3) Perform Fourier transform on the SAR block image to obtain the normalized power spectrum. According to the wavelength range of the internal wave, the power spectrum values ​​within the wavelength range are accumulated as the power spectrum eigenvalue.

[0047] (4) Setting the power spectrum feature threshold. When the feature value is greater than the threshold, it is roughly judged that the SAR image block contains internal waves, and the subsequent fine judgment and detection are carried out. If the feature value is less than the threshold, it is considered that the block does not contain internal waves, and the detection of the block is terminated.

[0048] (5) For the blocks roughly judged to contain internal waves, the elliptical domain normalized Radon transform is performed. The elliptical domain normalized Radon transform can eliminate the interference of diagonal effect and corner point effect on the straight line features in the image.

[0049] (6) Since there are many other phenomena in the ocean, such as swells, islands, shallow sea topography, oil pollution, marine mesoscale fronts, vortices, etc., which will interfere with the detection, it is necessary to improve the elliptical domain normalized Radon transform when extracting the Radon domain fronts and valleys corresponding to the wave lines of internal waves, so as to retain only the features corresponding to the wave lines of the ocean internal waves.

[0050] (7) According to the number and angle difference of the extracted Radon domain peaks, peak points or valleys, valley points, it is possible to further determine whether there are internal waves in the image area and eliminate false alarms.

[0051] (8) The Radon domain image that conforms to the distribution of internal wave lines is further processed to extract the internal wave direction according to the bending direction of the internal wave lines.

[0052] (9) Calculate the internal wave wavelength based on the distance between the front and front points or the valley and valley points in the Radon domain.

[0053] (10) The global density profile distribution feature database is loaded on board the satellite, the current block position is calculated according to the position obtained by the SAR image, the jump layer characteristics in the global profile database are queried in combination with the SAR acquisition time, the jump layer characteristics are combined with the internal wave wavelength characteristics, and the internal wave amplitude and wave velocity are calculated.

[0054] (11) Geographically mark the blocks containing internal waves and output the internal wave characteristic parameters, and transmit the onboard detection results to the ground.

[0055] 1. Internal wave power spectrum characteristics and threshold design:

[0056] By taking the two-dimensional Fourier transform of the image and integrating the two-dimensional Fourier transform energy according to the wave number, the wave number-power spectrum of the image can be obtained. The power spectrum can reflect the existence of stripes of a certain frequency in the image. When internal waves exist, the power spectrum characteristics of the wave number corresponding to the light and dark stripes of the internal wave will become larger. In this way, a rough step screening can be performed to determine whether there are internal wave stripes in the SAR image block. In this process, the extraction of power spectrum features and the setting of thresholds are the key to judging whether there are internal waves.

[0057] In order to better extract the characteristics of internal wave power spectrum, the ocean internal wave statistics were conducted on the previously collected Enviset, ERS-2, and Gaofen-3 satellite SAR data. A total of 167 ocean Enviset data were collected, including 24 ocean internal wave images, 232 ocean ERS-2 data, including 33 ocean internal wave images, and 67 ocean Gaofen-3 data, including 18 internal wave images. The internal wave wavelength was extracted by combining the traditional manual internal wave profile with the EMD method. The internal wave wavelength statistics results are shown in the figure below: Figure 2 shown.

[0058] Depend on Figure 2 It can be seen that the wavelength of ocean internal waves is mainly distributed in the range of 500 meters to 4000 meters, among which the number of internal waves with a wavelength of about 1500 meters is the largest.

[0059] When calculating the power spectrum of an image, the energy of the power spectrum area corresponding to the wavelength of the internal wave will increase due to the obvious light and dark stripes of the internal wave. By detecting the energy of the internal wave wavelength area, the internal wave can be detected to a certain extent. According to the distribution range of the typical internal wave wavelength of 500 meters to 3500 meters, the power spectrum value of the corresponding wave number can be integrated as the power spectrum characteristic parameter. If the image resolution is 50 meters, the sliding window is 512*512 pixels, and the area is about 25.6 square kilometers, when the wavelength of 500 meters to 3500 meters is integrated, the 7th to 51st points of the power spectrum wave number of the two-dimensional image are integrated.

[0060] 2. Improved algorithm of normalized Radon transform in elliptical domain and internal wave fringe positioning:

[0061] Due to the interference of other marine phenomena in the ocean, such as vortices, swells, oil films, ocean fronts and other factors, relying solely on power spectrum characteristics to detect internal wave areas will cause a large number of false alarms. Therefore, it is also necessary to consider the detailed characteristics of ocean internal waves to eliminate false alarms. Chen Jie et al. proposed using the elliptical domain normalized Radon transform method to detect ocean internal waves and extract internal wave characteristics. In actual application, this method is still subject to varying degrees of interference, resulting in the inability to extract the bright or dark spots corresponding to the light and dark lines of the internal waves in the Radon domain. Figure 3 , Figure 4 Two typical interference situations are given.

[0062] During on-board processing, in order to accurately detect and transmit the location of the block with internal waves, geo-reference will be performed before the on-board internal wave processing, and the land area will be removed, so the local data of the image is zero, such as Figure 3 As shown in (a), when the elliptical domain normalized Radon transform is used, the missing area causes two large dark areas and bright areas to appear in the Radon domain, such as Figure 3 As shown in (b), the detection of two points or dark points in the Radon domain is seriously disturbed, and the internal wave bright fringes and dark fringes corresponding to the bright and dark points cannot be detected normally.

[0063] Due to interference from eddies, swells, oil films, ocean fronts, etc. in the ocean, local areas in the image often become darker or brighter, such as Figure 4 (a) shows the image of ocean internal waves disturbed by oil film. At this time, the elliptical domain Radon transform is used. Affected by the interference area, a large dark area will also appear in the Radon domain, such as Figure 4 As shown in (b), the dark fringes corresponding to the internal waves cannot be detected correctly.

[0064] When it appears Figure 3 , Figure 4When there are two common interferences, the elliptical domain normalized Radon transform cannot correctly detect the bright or dark stripes of the internal wave, and the elliptical domain normalized Radon transform needs to be improved.

[0065] First, in order to remove the interference caused by missing image data, the missing part can be supplemented with the mean of the remaining part to avoid bright or dark areas caused by partial missing of the image during Radon transform, such as Figure 5 Shown Figure 3 The image is transformed in the elliptical domain by Radon transform after mean supplementation. It can be seen that after mean supplementation, the dark spots corresponding to the dark lines of the internal waves can be correctly found in the elliptical domain by Radon transform.

[0066] Secondly, the bright spots or dark spots corresponding to the bright and dark lines in the image can be obtained through the elliptical domain Radon transform, and the maximum and minimum areas can be obtained through the local maximum and local minimum methods. After the elliptical domain normalized Radon transform, in order to use a unified threshold to obtain the maximum and minimum points during onboard processing, the Radon transform image can be normalized again. For example, the Radon domain image can be stretched to the grayscale range of 0-255, and the extreme gradient domain value can be set according to the size of the image block. Taking the satellite image with a spatial resolution of 50 meters in the present invention and the internal wave detection block with a size of 512*512 pixels as an example, the extreme gradient domain value that can be set is 25. This parameter "25" can be set as an adjustable parameter and uploaded and adjusted according to needs. Figure 6 The following are shown: Figure 5 (a) and Figure 4 (a) Binary image after finding the minimum value.

[0067] pass Figure 6 (a) Minimum area, which can be roughly located Figure 5 (a) In the middle, there are dark fringes of internal waves, but Figure 6 There are several relatively large minimum regions in (b), which are mainly caused by Figure 4 The oil film interference in (a) is formed. The existence of these large-area interferences will also form large-area dark and bright areas in the Radon domain, and it is impossible to locate Figure 4 (a) The corresponding internal wave dark line, therefore, needs to be improved. After obtaining the binary map of the extreme value area of ​​the Radon domain, remove the area with a larger area. Taking the satellite image with a spatial resolution of 50 meters in the present invention and the internal wave detection block with a size of 512*512 pixels as an example, the area with an extreme value area exceeding 200 can be removed. This parameter "200" can be set as an adjustable parameter and uploaded and adjusted according to needs. Figure 7 Shown Figure 6 (b) Minimum binary image after removing large areas.

[0068] Internal solitary waves generally exist in the form of wave packets, and a wave packet contains multiple nearly parallel bright stripes, dark stripes, or light and dark stripes. According to the parallel characteristics of the stripes, it can be seen that in the Radon domain, the corresponding transformation angles are close to the same angle. Figure 3 The vertical angle of the internal wave stripes in (a) is about 180 degrees, so Figure 6 (a) In the Radon domain, there are multiple minimum points near the horizontal axis of 0 or 180, such as Figure 4 The dark stripes in (a) correspond to a vertical angle of about 37 degrees, so Figure 7 In the figure, the horizontal axis is around 37, and there are multiple minimum points. Through Radon transformation, the parallel straight lines in the detection image are transformed into extreme points with similar horizontal axes (angles) in the Radon domain. Therefore, to locate the internal wave fringes, it is only necessary to Figure 6 (a), Figure (7), in the improved Radon transform diagram, by finding the center point of the extreme value area close to the horizontal axis, the Radon domain extreme value point corresponding to the stripe in the image can be located. If there are more than two extreme value points close to the horizontal axis in the improved Radon domain, it is determined that there is an internal wave. If there is no extreme value point that meets the conditions, the area is determined to be a false alarm. Figure 8 Shown from Figure 6 (a) and Figure 7 The Radon domain image of the minimum point of the ocean that meets the internal wave conditions extracted from the ocean.

[0069] By improving the normalized Radon transform algorithm in the elliptical domain, the interference of missing data or other ocean phenomena in the original image can be effectively eliminated, and the maximum area of ​​the two internal wave stripes in the Radon domain and the minimum area of ​​the internal wave dark stripes in the Radon domain can be effectively found.

[0070] 3. Extraction of internal wave characteristic parameters:

[0071] Due to the special correspondence between Radon transform and image, the internal wave direction and wavelength information can be obtained by finding the relationship between the extreme points in the Radon domain. Combined with the jump layer information in the internal wave area, the internal wave amplitude and wave velocity can be finally inverted.

[0072] 3.1 Internal wave direction extraction:

[0073] Common internal solitary waves all have the characteristics that the leading wave is the longest and the waveguide line is bent toward the internal wave source. Based on this characteristic, the propagation direction of the internal wave is determined. According to the correspondence between the Radon transform and the original image, the maximum value of the extreme point that meets the internal wave condition is first calculated. For example, Figure 8The distribution image of the minimum points shown in the figure is used to find the location of the minimum value of these minimum areas. The connected minimum area where this minimum value is located should be the leading wave of the dark stripes of the internal wave in the original figure. The angle corresponding to the horizontal coordinate of the minimum value is the propagation direction of the internal wave, for example Figure 8 The minimum point in the minimum point block in (a) is at the coordinates (178, -135). Figure 8 In (b), the minimum point in the minimum point block is at the coordinate (37,175). In the Radon domain, the horizontal axis represents the Radon transform angle, and the vertical axis represents the distance from the center of the image. The propagation direction of the internal wave is consistent with the horizontal axis where the maximum point is located. Since the Radon transform has 180-degree ambiguity, the 180-degree ambiguity problem can be solved by the relationship between the center of gravity of the extreme value area where the maximum point is located and the position of the maximum value. Fig. 9 The figure shows the corresponding relationship between the internal wave propagation direction and the elliptical domain Radon domain. When the absolute value of the ordinate at the center of gravity of the region is greater than the absolute value of the ordinate at the extreme value of the region, it means that the internal wave line is curved inward, and the internal wave propagation direction should be propagating outward along the center point of the image. Therefore, the angle corresponding to the horizontal coordinate at this time is the propagation direction of the internal wave, such as Fig. 9 (a) is shown; when the absolute value of the ordinate where the center of gravity of the region is located is smaller than the absolute value of the ordinate where the extreme value of the region is located, it means that the internal wave line is bent outward, and the propagation direction of the internal wave should be along the outer side of the image toward the center point. At this time, the propagation direction of the internal wave is the angle corresponding to the horizontal coordinate plus 180 degrees, as shown in Fig. 9 (b).

[0074] Because the Radon transformation uses the Cartesian coordinate system, the right angle is 0 degrees, the north is 90 degrees, and the west is 180 degrees, which is inconsistent with the traditional geographical angle. Therefore, after obtaining the internal wave Radon domain angle, angle conversion is required. Assuming that the angle obtained in the Radon domain is θ, its coordinate is Θ under the geographical angle (the north is 0 degrees and the east is 90 degrees), then:

[0075]

[0076] 3.2 Calculation of internal wave wavelength

[0077] Yang Jinsong et al. used the KDV equation and SAR imaging principle to derive a SAR internal wave remote sensing simulation model for two-layer fluids, and obtained the relationship between the internal wave half-wave width L and the distance D between the brightest and darkest points of a single stripe in the SAR image:

[0078] D=1.32·L (2)

[0079] When the image has only light stripes or only dark stripes, the distance between two light stripes or the distance between two dark stripes can be used as 2D. In the Radon domain, the difference in ordinate between the maximum value and the second largest maximum value point or the difference in ordinate between the minimum value and the second smallest minimum value point that satisfies the internal wave condition can be obtained as 2D. Figure 8 (a) Calculated Figure 3 The corresponding internal wave dark fringe spacing is 53, according to Figure 8 (b) Calculated Figure 4 The corresponding internal wave dark fringe spacing is 76. According to the image spatial resolution, it can be calculated that Figure 3 The half-wave width of the inner wave is L = 1003.5m, Figure 4 The half-wave width of the middle inner wave is L=1439.4m.

[0080] 3.3 Calculation of internal wave amplitude and wave velocity:

[0081] Based on the dispersion relation of the internal wave KDV equation in the two-layer mode, a method for calculating the velocity and amplitude of internal waves from SAR images in a nonlinear framework is obtained.

[0082] Assume that in the two-layer mode, the thickness of the upper low-density water in the sea area is h1, the thickness of the lower water layer is h2, the density difference between the upper and lower water layers is Δρ, the average density of seawater is ρ, the gravitational acceleration is marked with g, the internal wave amplitude is η0, the internal wave phase velocity is C0, and the internal wave group velocity is C p , then the internal wave amplitude can be solved by equation (3), the internal wave phase velocity can be solved by equation (4), and the internal wave group velocity can be solved by equation (5).

[0083]

[0084]

[0085] The calculation of internal wave amplitude and wave velocity must be combined with the density profile of the sea area. According to the density profile, the thickness of the upper low-density water in the sea area is calculated to be h1, the thickness of the lower water layer is h2, the density difference between the upper and lower water layers is Δρ, and the average density of seawater is ρ. These data can be obtained through field measurements, but in general, it is difficult to synchronously measure the seawater density profile in the area where SAR obtains internal waves. In this case, the monthly average density profile data in WOD18 can be used to calculate these parameters.

[0086] 4. Test and results:

[0087] In order to test the effect of the method of the present invention on the automatic detection of internal waves in SAR ocean images, two typical images were selected for testing. Fig.10 (a) Fig.11 (a) shows a 50-meter resolution onboard SAR image acquired by a new satellite, which was georeferenced and de-landed onboard. Fig.10 (b) shows the method of the present invention. Fig.10 The internal wave area detected by the image, Fig.11 (b) shows the method of the present invention. Fig.11 (a) The internal wave region obtained by image detection. The sliding window size is 512 during detection. In order to avoid missed detection due to the splitting of internal waves by the sliding window, the sliding detection is performed by overlapping 256 pixels. The power spectrum rough detection threshold is set to 0.75, and the Radon domain extreme value detection gradient threshold is set to 25. Fig.10 , Fig.11 The inversion results of internal wave parameters are shown in Tables 1 and 2.

[0088] Table 1

[0089]

[0090] Table 2

[0091]

[0092]

[0093] 5. Conclusion: Internal wave on-board processing is subject to hardware resource limitations, and a logically concise, fast, and efficient detection method is required, while maintaining a high detection rate and a low false alarm rate as much as possible. From the results of automatic detection and parameter extraction of internal wave areas in two typical interfered SAR images, the areas where ocean internal waves are visually present in the images are all detected, and there are almost no false alarms. This shows that the method proposed in the present invention, which combines the image power spectrum features with the improved elliptical domain normalized Radon transform, can automatically detect ocean internal wave areas in SAR images, and quickly invert the wavelength, wave direction, amplitude, and wave velocity of internal waves. It has the advantages of high automation, high detection rate, and low false alarm rate, and can provide an important reference for automatic detection and feature inversion of ocean internal waves on-board SAR images.

[0094] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0096] In the description of the present invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0097] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A satellite inversion method for ocean internal wave SAR images, characterized in that: The following steps are involved: Based on the synthetic aperture radar SAR satellite, after correcting the image incident angle radiation non-uniformity, the sliding window size is set according to the image resolution and the typical spatial distribution range of the internal solitary wave, and each sliding window image block is processed in blocks; Perform Fourier transform on the SAR block image, obtain the normalized power spectrum, make a rough judgment, and obtain the block containing internal waves; For the blocks containing internal waves, the elliptical domain normalized Radon transform is performed to extract the Radon domain peaks and valleys corresponding to the wave lines of the internal waves, and further judge whether there are internal waves in the image area to eliminate false alarms; Geographically mark the blocks containing internal waves and output the internal wave characteristic parameters, and transmit the on-board detection results to the ground; In the process of extracting the Radon domain fronts and valleys corresponding to the wave lines of the internal waves, the elliptical domain normalized Radon transform is improved to obtain an improved Radon transform image, and the Radon domain fronts and valleys corresponding to the wave lines of the internal waves are extracted. The improvement process of the elliptical domain normalized Radon transform is as follows: after the missing part of the block containing the internal waves is supplemented with the mean value of the remaining part, the bright spots or dark spots corresponding to the bright lines and dark lines of the image are obtained by using the elliptical domain Radon transform, and the maximum value area and the minimum value area are obtained according to the local maximum value and local minimum value method. When processing on the satellite, a unified threshold is used to obtain the maximum value and the minimum value points, and the Radon transform image is normalized again; In the process of further judging whether there is an internal wave in the image region, a further judgment is made on whether there is an internal wave in the image region according to the number and angle difference of the extracted Radon domain fronts, front points or valleys, valley points; In the process of eliminating false alarms, the internal wave stripes are located, and in the improved Radon transform diagram, the center point of the extreme value area close to the horizontal axis is obtained, that is, the Radon domain extreme value point corresponding to the stripes in the positioning image is located. If there are more than two extreme value points close to the horizontal axis in the improved Radon domain, it is determined that there are internal waves. If there are no extreme value points that meet the conditions, the area is determined to be a false alarm, and the false alarm is eliminated.

2. The on-board inversion method for ocean internal wave SAR images according to claim 1, characterized in that: In the process of performing block processing on each sliding window image block, based on the set sliding window size, sliding is performed in a partially overlapping manner to perform block processing on each sliding window image block.

3. The on-board inversion method for ocean internal wave SAR images according to claim 2, characterized in that: In the process of rough judgment, according to the wavelength range of internal waves, the power spectrum values ​​within the wavelength range are accumulated as the power spectrum characteristic value. By setting the power spectrum characteristic threshold, it is roughly judged that the SAR block image contains internal waves.

4. The on-board inversion method for ocean internal wave SAR images according to claim 1, characterized in that: In the process of obtaining the internal wave characteristic parameters of the block containing the internal wave, the internal wave characteristic parameters include the internal wave direction, the internal wave wavelength, and the internal wave amplitude and wave velocity.

5. The on-board inversion method for ocean internal wave SAR images according to claim 4, characterized in that: In the process of extracting the internal wave direction, the internal wave direction is extracted based on the bending direction of the internal wave line of the Radon domain image that conforms to the internal wave line distribution.

6. The on-board inversion method for ocean internal wave SAR images according to claim 5, characterized in that: In the process of obtaining the internal wave wavelength, the internal wave wavelength is obtained according to the distance between the front and the front point or the valley and the valley point in the Radon domain.

7. The on-board inversion method for ocean internal wave SAR images according to claim 6, characterized in that: In the process of obtaining the internal wave amplitude and velocity, the current block position is calculated according to the position of the SAR image acquisition, and the jump layer characteristics are obtained in combination with the SAR acquisition time. The jump layer characteristics are combined with the internal wave wavelength characteristics to calculate the internal wave amplitude and velocity.

Images