A method and system for detecting ocean vortex in SAR images

Abstract

This invention discloses a method and system for detecting ocean eddies in SAR images. Starting from the range Doppler domain, it fully considers the velocity characteristics of ocean eddies. Using sub-images formed by segmentation of SAR single-view complex images as the analysis object, it calculates the Doppler velocity gradient of the sea surface to determine the eddy center and eddy scale. This effectively improves the detection probability of ocean eddies based on SAR images and reduces the false detection probability of ocean eddies based on SAR images. The system is simple and the method is feasible.

Description

Technical Field

[0001] The invention belongs to the technical field of synthetic aperture radar (SAR) signal processing and the technical field of ocean remote sensing application, and in particular relates to a SAR image ocean vortex detection method and system. Background Art

[0002] Ocean vortex is a rotating body of water with closed circulation as its main feature. It is caused by various meteorological factors and ocean dynamic instability. It is widely distributed in the world's oceans and plays an important role in the ocean and atmospheric environment. Synthetic Aperture Radar (SAR) is an active microwave imaging sensor. Compared with traditional optical remote sensing and infrared remote sensing, SAR has all-weather, all-day, large-scale, high-resolution imaging capabilities, which gives it a unique advantage in marine applications and has become one of the main means of detecting ocean vortices.

[0003] The existing ocean vortex detection algorithm of SAR images is mainly based on the light and dark features of ocean vortices on SAR amplitude maps, that is, it uses the grayscale gradient features of ocean vortices for detection. The grayscale of SAR images mainly depends on the sea surface scattering intensity, and the mechanism of changing the sea surface scattering intensity is mainly the enhancement or weakening of Bragg waves caused by the "convergence and divergence" of sea surface flow. This "convergence and divergence" effect mainly depends on the divergence of the sea surface flow field. However, for ocean vortices, the sea surface flow field is mainly curled, and the divergence is very weak, making it difficult to form significant light and dark textures on SAR images; on the other hand, the brightness and darkness of ocean vortices are usually affected by multiple factors such as radar imaging systems and sea conditions, making some ocean vortices show very weak light and dark features in SAR amplitude images, or even invisible, which brings great trouble to the detection algorithm based on amplitude. Therefore, in order to accurately detect ocean vortices from SAR images, it is urgent to explore other features of ocean vortices so that the detection of ocean vortices is not limited to SAR amplitude maps.

[0004] In summary, the existing ocean eddy detection methods in SAR images mainly start from the amplitude and only consider the grayscale features, which inevitably leads to high false alarm rates and missed detection rates. Summary of the invention

[0005] In view of this, the present invention provides a SAR image ocean vortex detection method and system, which can detect ocean vortex based on sea surface Doppler velocity.

[0006] The present invention provides a SAR image ocean vortex detection method, comprising:

[0007] Step 1, dividing the SAR single-view complex image into multiple sub-image blocks;

[0008] Step 2, estimating the Doppler center frequency of the sub-image block as the observed value of the Doppler center frequency;

[0009] Step 3: Calculate the predicted value of the Doppler center frequency of the sub-image block based on the satellite orbit parameters and the antenna pointing parameters;

[0010] Step 4: The difference between the observed value and the predicted value of the sub-image block is the Doppler center outlier value of the sub-image block, and the Doppler center outlier value is converted into the sea surface Doppler velocity;

[0011] Step 5, calculating the sea surface Doppler velocity gradient to form a sea surface Doppler velocity gradient matrix, and determining the maximum value of the absolute value of the elements in the sea surface Doppler velocity gradient matrix as the vortex center;

[0012] Step 6: Determine the vortex edge according to the variation law of the Doppler velocity of the sea surface in the vicinity of the vortex center, and calculate the vortex scale.

[0013] Furthermore, the process of dividing the SAR single-view complex image into multiple sub-image blocks in step 1 is: the SAR single-view complex image is evenly divided along the azimuth and range directions to obtain M*N sub-image blocks, wherein M and N are both positive integers not less than 3.

[0014] Furthermore, in step 2, the method of estimating the Doppler center frequency of the sub-image block as the observation value of the Doppler center frequency is: estimating the observation value of the Doppler center frequency of the sub-image block by using a time domain autocorrelation method.

[0015] Furthermore, the SAR single-view complex image is obtained by processing using RD, CS or Omega-K algorithm.

[0016] The present invention provides a SAR image ocean vortex detection system, comprising:

[0017] An image segmentation module is used to segment the SAR single-view complex image into M*N sub-image blocks;

[0018] An observation value estimation module, used for estimating the Doppler center frequency of the sub-image block to obtain an observation value of the Doppler center frequency;

[0019] A prediction value calculation module, used for calculating the prediction value of the Doppler center frequency of the sub-image block based on the satellite orbit parameters and the antenna pointing parameters;

[0020] an outlier value calculation module, used for calculating the Doppler center outlier value of the sub-image block based on the observed value of the Doppler center frequency output by the observed value estimation module and the predicted value of the Doppler center frequency output by the predicted value calculation module, and then converting the Doppler center outlier value into the sea surface Doppler velocity;

[0021] The ocean vortex detection module is used to calculate the sea surface Doppler velocity gradient, determine the point with the maximum absolute value of the elements in the sea surface Doppler velocity gradient matrix as the vortex center; and then determine the vortex edge and calculate the vortex scale according to the change law of the sea surface Doppler velocity in the vortex center and its neighborhood.

[0022] Beneficial effects:

[0023] The present invention starts from the range Doppler domain, fully considers the velocity characteristics of the ocean vortex, takes the sub-image formed by segmenting the SAR single-view complex image as the analysis object, calculates the sea surface Doppler velocity gradient, thereby determining the vortex center and vortex scale, effectively improving the ocean vortex detection probability based on the SAR image, and reducing the ocean vortex false detection probability based on the SAR image. The system is simple and the method is feasible. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The present invention provides a flow chart of a SAR image ocean vortex detection method.

[0025] Figure 2 A detailed flow chart of a SAR image ocean vortex detection method provided by the present invention.

[0026] Figure 3 The invention discloses a single-view complex image obtained by imaging ENVISAT ASAR data selected by a SAR image ocean eddy detection method provided by the present invention.

[0027] FIG. 4( a ) is a Doppler center anomaly map of a single-view complex image obtained by using a SAR image ocean vortex detection method provided by the present invention.

[0028] FIG. 4( b ) is a single-view complex image block sea surface Doppler velocity map obtained by using a SAR image ocean vortex detection method provided by the present invention.

[0029] FIG. 5( a ) is a diagram showing the detection results of the center and edge of an ocean vortex on a sea surface Doppler velocity map obtained by using a SAR image ocean vortex detection method provided by the present invention.

[0030] FIG5( b ) is a diagram showing the detection results of the ocean vortex center and edge mapped on a single-view complex image obtained by using a SAR image ocean vortex detection method provided by the present invention.

[0031] FIG. 6( a ) is a projection of an ocean vortex center and edge detection result image obtained by using a SAR image ocean vortex detection method provided by the present invention on a Google map.

[0032] FIG6( b ) is a verification diagram of an ocean vortex detection result obtained by using a SAR image ocean vortex detection method provided by the present invention. DETAILED DESCRIPTION

[0033] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0034] Traditional SAR image ocean eddy detection methods are usually based on SAR grayscale image processing only. The grayscale of SAR images mainly depends on the sea surface scattering intensity, and the mechanism of changing the sea surface scattering intensity is mainly the Bragg wave enhancement or weakening caused by the "convergence and divergence" of the sea surface flow. This "convergence and divergence" effect mainly depends on the divergence of the sea surface flow field. However, for ocean eddies, the sea surface flow field is mainly curled, and the divergence is very weak, making it difficult to form significant light and dark textures on SAR images.

[0035] Ocean vortices are ubiquitous in the ocean, and their huge kinetic energy can cause strong seawater movement in local areas of the ocean. Their movement speed is several times or even an order of magnitude faster than the average flow speed of ocean currents, and the SAR echo phase contains rich target motion information, which is difficult to obtain from SAR amplitude images. The huge velocity gradient at the center of the ocean vortex and the good symmetry of the velocity direction on both sides of the vortex center are two important characteristics of the vortex. Based on the SAR sea surface single-view complex image data, the SAR echo phase information is used to extract the above two features, thereby realizing the detection of the invisible ocean vortex center and vortex edge in the amplitude image, and the echo data detection results are complemented by the amplitude detection results to extract richer vortex information and realize more accurate detection of the vortex. Based on the above analysis, the present invention proposes a SAR image ocean vortex detection method and system, and the process is as follows: Figure 1 shown.

[0036] The present invention provides a SAR image ocean vortex detection method, the detailed process is as follows: Figure 2 As shown, the specific steps include:

[0037] Step 1: Divide the SAR single-view complex image into M*N sub-image blocks.

[0038] Generally, SAR single-view complex images can be obtained through SAR imaging algorithms such as RD, CS, and Omega-K, or they can be downloaded from public data websites.

[0039] The SAR single-view complex image is evenly divided into M*N sub-image blocks along the azimuth and range directions, and the size of each sub-image block is the same. The number of divisions in the azimuth direction is M, and the number of divisions in the range direction is N, where M and N are both positive integers not less than 3. At the same time, considering the spatial variation characteristics of Doppler data along the azimuth and range directions, the velocity characteristics of ocean phenomena of different scales, and the influence of noise, the size of each sub-image block needs to be determined according to needs, and the corresponding spatial resolution is not less than 0.5km*0.5km.

[0040] Step 2: Estimate the Doppler center frequency of each sub-image block as the observed value of the Doppler center frequency.

[0041] Estimate the Doppler center frequency of each sub-image block. Since each sub-image block of the SAR single-view complex image can be regarded as a stationary complex Gaussian process, according to the Wiener-Khinchin theorem, the method based on time domain autocorrelation can be used to estimate the observed value f of the Doppler center frequency of each sub-image block. Dc , the formula used is as follows:

[0042]

[0043] Among them, arg{·} is the angle to be calculated, r(·) is the correlation function, PRF is the pulse repetition frequency, T is the sampling interval and T = 1 / PRF. Usually, when the value of k is 1, the estimation method has strong robustness. At this time, the autocorrelation function r(T) of each sub-image block can be expressed as shown in the following formula:

[0044]

[0045] Among them, s(m,n) is the (m,n)th complex value of the sub-image block, Na is the number of pixels in the azimuth direction of the sub-image block, and Nr is the number of pixels in the distance direction of the sub-image block. At this time, the observed value f of the Doppler center frequency of each sub-image block is estimated based on the time domain autocorrelation method Dc , as shown below:

[0046]

[0047] Step 3: Calculate the predicted value of the Doppler center frequency of each sub-image block based on the satellite orbit parameters and the antenna pointing parameters.

[0048] The system factors that affect the Doppler frequency can be summarized as the satellite orbit geometry model, the radar beam pointing, and the intersection of the beam and the rotating surface. This part of the Doppler is called the predicted value of the Doppler center frequency. The predicted value of the Doppler center frequency of each sub-image block is estimated based on the satellite orbit parameters and antenna pointing parameters. Dp Use the following formula:

[0049]

[0050] Among them, k e is the radar electromagnetic wave number, V sc is the speed of the radar platform along the track direction, γ is the altitude angle of the radar beam, α is the yaw angle, ω e is the angular velocity of the Earth's rotation, ω is the angular velocity of the radar platform's orbital rotation, ε is the radar's side-view direction (e.g., +1 indicates right-side viewing, -1 indicates left-side viewing), β is the angle between the ascending node and the satellite's orbital plane, and ψ is the satellite's orbital inclination.

[0051] Step 4: Calculate the Doppler center outlier value of each sub-image block according to the observed value and predicted value of the Doppler center frequency, and convert the obtained Doppler center outlier value into the sea surface Doppler velocity.

[0052] The Doppler center frequency observation value f of the sub-image block is Dc and the predicted value of Doppler center frequency f Dp The Doppler center outlier value f can be obtained by subtracting Dca , as shown in the following formula:

[0053] f Dca =f Dc -f Dp (5).

[0054] The Doppler center outlier reflects the speed of the surface object. The Doppler center outlier is a vector whose sign represents the direction of the object's movement relative to the radar's line of sight. The sign is positive when it is close to the radar and negative when it is far away from the radar. Its absolute value represents the magnitude of the object's speed. Generally, for an object that is stationary relative to the ground, its Doppler center outlier is zero.

[0055] According to the Doppler center outlier, the velocity of the object relative to the radar radial direction can be obtained, that is, the sea surface Doppler velocity, the sea surface Doppler velocity v D The calculation formula is as follows:

[0056]

[0057] Among them, θ i is the incident angle of the radar beam relative to the sea surface.

[0058] Step 5: Calculate the sea surface Doppler velocity gradient to obtain the sea surface Doppler velocity gradient matrix, and determine the maximum absolute value of the elements in the sea surface Doppler velocity gradient matrix as the vortex center.

[0059] Since the original single-view complex image is divided into M*N sub-image blocks, the sea surface Doppler velocity v D This forms an M*N dimensional matrix:

[0060] v D =[v D1 ,v D2 ,···,v Di ,···,v DN ](7)

[0061] Among them, v Di (i=1,2,···,N) is an M-dimensional column vector.

[0062] Find the sea surface Doppler velocity matrix v D The gradient of the sea surface Doppler velocity gradient matrix is ​​obtained:

[0063] ▽v D =(▽v Dij ) M×N (8)

[0064] Take the sea surface Doppler velocity gradient matrix ▽v D The coordinates corresponding to the maximum absolute value of the elements in are:

[0065]

[0066] Coordinate A is the coordinate of the vortex center.

[0067] Step 6: Determine the vortex edge according to the variation law of the Doppler velocity of the sea surface in the vicinity of the vortex center, and calculate the vortex scale.

[0068] The vortex edge is determined according to the variation law of the Doppler velocity of the sea surface in the vortex center coordinate A and its neighborhood, and the vortex scale is calculated. The Doppler velocity of the sea surface in the vortex center coordinate and its neighborhood is clustered, and the Doppler velocity gradient is obtained at the same time. The gradient mutation point is found as the vortex edge, and the distance from all edges to the vortex center is calculated. The edge corresponding to the maximum distance is taken as the final vortex edge position, and the vortex scale is calculated on this basis.

[0069] The invention provides a SAR image ocean vortex detection system, which comprises an image segmentation module, an observation value estimation module, a prediction value calculation module, an abnormal value calculation module and an ocean vortex detection module.

[0070] The image segmentation module is used to segment the SAR single-view complex image into M*N sub-image blocks.

[0071] The observation value estimation module is used to estimate the Doppler center frequency of the sub-image block and use the estimation result as the observation value of the Doppler center frequency.

[0072] The prediction value calculation module is used to calculate the prediction value of the Doppler center frequency of the sub-image block based on the satellite orbit parameters and the antenna pointing parameters.

[0073] The outlier value calculation module is used to calculate the Doppler center outlier value of the sub-image block based on the observed value of the Doppler center frequency output by the observed value estimation module and the predicted value of the Doppler center frequency output by the predicted value calculation module, and convert the obtained Doppler center outlier value into the sea surface Doppler velocity.

[0074] The ocean vortex detection module is used to calculate the sea surface Doppler velocity gradient and determine the point with the maximum absolute value of the elements in the sea surface Doppler velocity gradient matrix as the vortex center. The vortex edge is determined according to the variation law of the sea surface Doppler velocity in the vortex center and its neighborhood, and the vortex scale is calculated.

[0075] Example:

[0076] This embodiment uses a SAR image ocean vortex detection method provided by the present invention to detect ocean vortices, including the following steps:

[0077] First, SAR raw data is acquired and imaged to generate a single-view complex image. In this embodiment, ENVISAT ASAR single-view complex images are selected to detect ocean vortices. Figure 3 Shown is a strip SAR data located near the Luzon Islands. The data was acquired on August 26, 2008, in an ascending orbit. The vortex is almost invisible from the HH polarization SAR amplitude map.

[0078] Then, the acquired single-view complex image is subjected to azimuth Fourier transform to obtain range Doppler domain data, which is averagely segmented in azimuth and range to obtain sub-image blocks for subsequent Doppler estimation. In this embodiment, the azimuth sub-image block M is 60, and the range sub-image block N is 51. The Doppler center anomaly of each sub-image block is estimated according to formulas (1)-(5), as shown in FIG4(a), and the Doppler center anomaly value is converted into the sea surface Doppler velocity according to formula (6), as shown in FIG4(b).

[0079] Next, the sea surface Doppler velocity gradient is calculated for each sub-image block according to formula (8), and the absolute maximum value of the element in each sea surface Doppler velocity gradient matrix, that is, the location of the vortex center, is determined according to formula (9). The vortex edge is determined according to the law of sea surface Doppler velocity change in the neighborhood of the vortex center, and the result is shown in Figure 5(a).

[0080] Finally, the vortex center and edge detection results shown in Figure 5(a) are mapped to the original single-view complex image, as shown in Figure 5(b).

[0081] In order to further verify the correctness of the results, MODIS Chl-a (chlorophyll-a) data was selected for comparative analysis. The MODIS Chl-a data is shown in Figure 6(b), and the data acquisition time is 04:45:07 UTC on August 26, 2008. Figure 6(a) is the projection of the SAR image on the Google map. According to the bright and dark characteristics of the vortex in the SAR image, the detection result is reasonable. From the comparison of the results in Figure 6(a) and Figure 6(b), it can be seen that the mesoscale ocean vortex detection results based on SAR echoes are roughly the same as those of MODIS Chl-a data. Since there is a 9-hour deviation between the SAR data vortex acquisition time and the MODIS data vortex acquisition time, there is a certain offset between the vortex positions of the two data, but the vortex scales detected by the two data are roughly the same, which further proves the effectiveness of the algorithm.

[0082] At this point, the detection of ocean eddies in SAR images based on sea surface Doppler velocity has been completed.

[0083] In summary, the above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting ocean vortexes using SAR images, characterized in that: include: Step 1, dividing the SAR single-view complex image into multiple sub-image blocks; Step 2, estimating the Doppler center frequency of the sub-image block as the observed value of the Doppler center frequency; Step 3: Calculate the predicted value of the Doppler center frequency of the sub-image block based on the satellite orbit parameters and the antenna pointing parameters; Step 4: The difference between the observed value and the predicted value of the sub-image block is the Doppler center outlier value of the sub-image block, and the Doppler center outlier value is converted into the sea surface Doppler velocity; Step 5, calculating the sea surface Doppler velocity gradient to form a sea surface Doppler velocity gradient matrix, and determining the maximum value of the absolute value of the elements in the sea surface Doppler velocity gradient matrix as the vortex center initial value; Step 6: Starting from the maximum value of the absolute value of the Doppler velocity gradient, analyze the Doppler velocity in its neighborhood to determine whether it meets the vortex characteristics; if it meets the vortex characteristics, it is determined as the final vortex center and proceeds to step 7; if it does not meet the requirements, find the next largest value again and repeat this step until the vortex conditions are met; Step 7: Cluster the coordinates of the vortex center and the Doppler velocity of the sea surface in its neighborhood, and obtain the Doppler velocity gradient at the same time, find the gradient mutation point as the vortex edge, calculate the distance from all edges to the vortex center, take the position corresponding to the maximum distance as the final vortex edge, and calculate the vortex scale on this basis.

2. A SAR image ocean vortex detection method according to claim 1, characterized in that: The process of dividing the SAR single-view complex image into multiple sub-image blocks in step 1 is: the SAR single-view complex image is evenly divided along the azimuth and range directions to obtain M*N sub-image blocks, wherein M and N are both positive integers not less than 3.

3. A SAR image ocean vortex detection method according to claim 1, characterized in that: The method of estimating the Doppler center frequency of the sub-image block as the observation value of the Doppler center frequency in step 2 is: estimating the observation value of the Doppler center frequency of the sub-image block by using a time domain autocorrelation method.

4. A SAR image ocean vortex detection method according to claim 1, characterized in that: The SAR single-view complex image is obtained by processing with RD, CS or Omega-K algorithm.

5. A SAR image ocean vortex detection system based on the method of claim 1, characterized in that: include: An image segmentation module is used to segment the SAR single-view complex image into M*N sub-image blocks; An observation value estimation module, used for estimating the Doppler center frequency of the sub-image block to obtain an observation value of the Doppler center frequency; A prediction value calculation module, used for calculating the prediction value of the Doppler center frequency of the sub-image block based on the satellite orbit parameters and the antenna pointing parameters; an outlier value calculation module, used for calculating the Doppler center outlier value of the sub-image block based on the observed value of the Doppler center frequency output by the observed value estimation module and the predicted value of the Doppler center frequency output by the predicted value calculation module, and then converting the Doppler center outlier value into the sea surface Doppler velocity; The ocean vortex detection module is used to calculate the sea surface Doppler velocity gradient, determine the point with the maximum absolute value of the elements in the sea surface Doppler velocity gradient matrix as the vortex center; and then determine the vortex edge and calculate the vortex scale according to the change law of the sea surface Doppler velocity in the vortex center and its neighborhood.

Images