A method, device and storage medium for segmenting oceanic internal wave ripples in SAR images

Abstract

The application relates to a SAR image ocean internal wave stripe segmentation method, equipment and a storage medium, the method comprises the following steps: step S1, acquiring an ocean internal wave SAR image, and performing feature extraction on the ocean internal wave SAR image by adopting Gabor transformation to obtain a feature image; step S2, stripe segmentation is performed on the feature image by using a K-means clustering algorithm, the bright / dark stripes in the ocean internal wave are separated from the background, and the regions where the dark / bright stripes are located are automatically judged according to the inter-class difference; step S3, a minimum circumscribed rectangle of the bright / dark stripe contour is constructed, and the dark / bright stripe position is obtained by translating a set distance along the normal direction of the long side of the minimum circumscribed rectangle. Compared with the prior art, the method has better bright / dark stripe segmentation effect on the SAR image with strong ocean internal wave characteristics, can better identify the bright or dark stripe features with large differences from the background, and has the advantages of strong robustness and high accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of internal wave detection, in particular to a SAR image ocean internal wave stripe segmentation method, device and storage medium. BACKGROUND

[0002] Ocean internal wave is a kind of wave phenomenon existing in the stable stratified ocean. It has the characteristics of strong burst, strong destructive power and huge energy. Ocean internal wave has a great influence on the dynamic process of ocean structure and surface, and also seriously affects the safety of ocean engineering structure. Therefore, it is necessary to accurately understand the position of ocean internal wave occurrence.

[0003] Due to the advantages of all-weather, all-day and high-resolution imaging of synthetic aperture radar (SAR), the remote sensing research of ocean internal wave has attracted widespread attention. However, in the existing remote sensing research of ocean internal wave, only the information of ocean internal wave stripe is extracted, and the specific positions of light and dark stripes are not distinguished. Due to the strong subjectivity and low efficiency of naked eye observation, it is urgent to extract features from images to reflect the evolution rule and change process of things.

[0004] The light and dark stripe characteristics of ocean internal wave can directly reflect the propagation information and evolution rule of ocean internal wave, so it is of great significance to extract features of ocean internal wave in SAR image and separate the light and dark stripes of ocean internal wave from the background. SUMMARY

[0005] The purpose of the present application is to overcome the defects of the prior art and provide a SAR image ocean internal wave stripe segmentation method, device and storage medium with good light and dark stripe segmentation effect.

[0006] The purpose of the present application can be achieved by the following technical solutions:

[0007] According to the first aspect of the present application, a SAR image ocean internal wave stripe segmentation method is provided, which comprises the following steps:

[0008] Step S1, acquiring ocean internal wave SAR image, and using Gabor transform to extract features of the ocean internal wave SAR image to obtain a feature image;

[0009] Step S2, using K-means clustering algorithm to segment the stripes of the feature image, separating the light stripes from the background in the ocean internal wave, and automatically judging the region of dark stripes according to the inter-class difference, or separating the dark stripes from the background in the ocean internal wave, and automatically judging the region of light stripes according to the inter-class difference;

[0010] Step S3, based on the area where the dark stripes are located, constructing the minimum circumscribed rectangle of the bright stripe profile, translating a set distance along the normal direction of the long side of the minimum circumscribed rectangle to obtain the location of the dark stripes, or based on the area where the bright stripes are located, constructing the minimum circumscribed rectangle of the dark stripe profile, translating a set distance along the normal direction of the long side of the minimum circumscribed rectangle to obtain the location of the bright stripes.

[0011] Preferably, in the step S1, the Gabor transform is used to extract features of the ocean internal wave SAR image to obtain a feature image, specifically: based on the corresponding relationship between the Gabor kernel function and the filter image, a Gabor kernel function representing different wavelengths and directions is constructed, which is respectively convolved with the ocean internal wave SAR image to obtain a multi-dimensional filter image responding to different features under different Gabor kernel functions, and then the multi-dimensional filter image is synthesized into a one-dimensional image through Gaussian smoothing to extract a feature image for classification.

[0012] Preferably, the Gabor transform is an even-symmetry filter taking only the real part of the Gabor filter, and the expression of the corresponding real part of the Gabor filter is:

[0013]

[0014] In the formula, x and y are pixel position coordinates, x' and y' are new coordinates generated by rotating the Gabor kernel function, λ is the wavelength of the cosine function; θ is the normal direction of the parallel strip, i.e. the texture direction, which is selected at a set interval between 0°-180°; ψ is the phase parameter of the cosine function, and ψ is 0°; σ is the standard deviation of the Gaussian function; γ is the space aspect ratio of the Gabor function.

[0015] Preferably, the corresponding relationship between the wavelength λ of the cosine function and the spatial frequency is:

[0016]

[0017] In the formula, the spatial frequency is transformed to between -0.5 and 0.5 through normalization, F L is the spatial frequency between 0 and 0.25, F H is the spatial frequency between 0.25 and 0.5; N c is the image size; σ is obtained from b is the half-response spatial frequency bandwidth.

[0018] Preferably, the texture direction θ in the step S1 is selected at an interval of 30° between 0°-180°.

[0019] Preferably, the Gaussian function expression used in the Gaussian smoothing is:

[0020]

[0021] wherein σ G is the standard deviation of the Gaussian window, and x, y are the pixel position coordinates.

[0022] Preferably, the step S2 uses a K-means clustering algorithm to segment the feature image, specifically as follows:

[0023] 1) initialize the number of clusters K, randomly create an initial partition, and give the initial cluster centers;

[0024] 2) use an iterative method to improve the partition effect by constantly moving the cluster centers until the criterion function converges, wherein the criterion function expression is:

[0025]

[0026] wherein E is the sum of squared errors, k is the number of categories, x is the current category value, is the average value of category C i .

[0027] Preferably, the step S3 constructs a minimum circumscribed rectangle of the dark stripe contour based on the area where the bright stripe is located, and translates a set distance along the normal direction of the long side of the minimum circumscribed rectangle to obtain the location of the bright stripe, specifically as follows:

[0028] Based on the characteristics of the ocean internal wave image, the minimum circumscribed rectangle of each dark stripe contour is used to translate a set distance D i along the normal direction of the long side of the rectangle, which is the location of each bright stripe; the segmented dark stripe region is denoted as set m, and the region is translated a distance D i along the normal direction of the inside and outside, respectively, to obtain two regions of the same size, denoted as n and p, forming three regions of the same size, i.e., the same number of coordinate point sets; wherein set m corresponds to the dark stripe, set n corresponds to the bright stripe or the background region, and set p is determined accordingly; in the original image, the side with a larger inter-class difference is automatically determined as the bright stripe region, and the bright stripe is automatically extracted

[0029] According to a second aspect of the present application, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement any of the methods.

[0030] According to a third aspect of the present application, a computer readable storage medium is provided, which stores a computer program, and the program is executed by a processor to implement any of the methods.

[0031] Compared with the prior art, the present application has the following advantages:

[0032] 1) The research of the present application has better segmentation effect on the bright and dark stripes of the SAR image with strong internal wave characteristics in the sea, and can better identify the bright or dark stripe characteristics with large difference from the background.

[0033] 2) The present application can further automatically determine the relative position of the corresponding dark or bright stripe by comparing the inter-class difference, thereby realizing the feature extraction of the bright and dark stripes of the internal wave in the sea, and laying a foundation for the subsequent research on the parameter inversion of the internal wave in the sea. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 The SAR image (11200x9600) of the internal wave in the sea (the central point longitude and latitude coordinates are 48°35'W, 29°17'S);

[0035] Figure 2 The SAR image (1024x1024) of the internal wave in the selected area;

[0036] Figure 3 The flowchart of the feature extraction of the internal wave in the sea;

[0037] Figure 4 The segmentation effect diagram of the dark stripe under the optimal IoU;

[0038] Figure 5 The segmentation effect comparison diagram of the dark stripe under the optimal IoU;

[0039] Figure 6 The stripe translation schematic diagram;

[0040] Figure 7 The bright and dark stripe segmentation diagram of the internal wave in the sea; wherein a, b and c are three different SAR images of the internal wave in the sea; i-vi are the original image of the SAR image of the internal wave in the sea, the manually segmented image, the segmented image after the median filtering, the segmented image after the Gabor filtering, the bright and dark stripe position diagram, and the bright and dark stripe comparison diagram;

[0041] Figure 8 The flowchart of the method of the present application. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.

[0043] The present application provides a SAR image internal wave stripe segmentation method, as shown in Figure 8 The method comprises the following steps:

[0044] Step S1, acquire the ocean internal wave SAR image, and use Gabor transform to extract features of the ocean internal wave SAR image to obtain a feature image, specifically:

[0045] Based on the corresponding relationship between the Gabor kernel function and the filtered image, the Gabor kernel function representing different wavelengths and directions is constructed, which is respectively convolved with the ocean internal wave SAR image to obtain multi-dimensional filtered images responding to different features under different Gabor kernel functions, and then the multi-dimensional filtered images are synthesized into one-dimensional images through Gaussian smoothing to extract the feature images for classification;

[0046] The Gabor transform is an even-symmetry filter taking only the real part of the Gabor filter, and the expression of the corresponding real part of the Gabor filter is:

[0047]

[0048] In the formula, x and y are pixel position coordinates, x' and y' are new coordinates generated by rotating the Gabor kernel function, λ is the wavelength of the cosine function; θ is the normal direction of the parallel strip, i.e. the texture direction, which is selected at a set interval between 0°-180°, and in the embodiment, the texture direction θ is selected at an interval of 30° between 0°-180°; ψ is the phase parameter of the cosine function, and the value of ψ is 0°; σ is the standard deviation of the Gaussian function; γ is the space aspect ratio of the Gabor function.

[0049] The corresponding relationship between the wavelength λ of the cosine function and the spatial frequency is:

[0050]

[0051] In the formula, the spatial frequency is transformed to between-0.5 and 0.5 through normalization, F L is the spatial frequency between 0 and 0.25, F H is the spatial frequency between 0.25 and 0.5; N c is the image size; σ is obtained from , b is the half-response spatial frequency bandwidth, and in the embodiment, the default value 2 is taken.

[0052] The Gaussian function expression used in Gaussian smoothing is:

[0053]

[0054] In the formula, σ G is the Gaussian window standard deviation, which is set to σ G = 3σ in the embodiment; x and y are pixel position coordinates.

[0055] Step S2: Use the K-means clustering algorithm to perform stripe segmentation on the feature image, separate the bright stripes in the ocean internal wave from the background, and automatically determine the region where the dark stripes are located based on the inter-class difference, or separate the dark stripes in the ocean internal wave from the background and automatically determine the region where the bright stripes are located based on the inter-class difference.

[0056] The K-means clustering algorithm is used to segment the stripes in the feature image, specifically as follows:

[0057] 1) Initialize the number of clusters K, randomly create an initial partition, and provide the initial cluster centers;

[0058] 2) An iterative method is used to improve the partitioning effect by continuously moving the cluster centers until the criterion function converges. The expression for the criterion function is:

[0059]

[0060] In the formula, E is the sum of squared errors, k is the number of categories, and x is the current category value. Category C i The average value.

[0061] Step S3: Based on the region where the dark stripe is located, construct the minimum bounding rectangle of the bright stripe outline, and translate it by a set distance along the normal direction of the long side of the minimum bounding rectangle to obtain the location of the dark stripe; or based on the region where the bright stripe is located, construct the minimum bounding rectangle of the dark stripe outline, and translate it by a set distance along the normal direction of the long side of the minimum bounding rectangle to obtain the location of the bright stripe.

[0062] Based on the region where the bright fringe is located, construct the minimum bounding rectangle of the dark fringe outline. Translate the rectangle by a set distance along the normal direction of its long side to obtain the location of the bright fringe. Specifically:

[0063] Based on the characteristics of ocean internal wave images, the minimum bounding rectangle of each dark fringe contour is used, and the image is translated by a set distance D along the normal direction of the long side of the rectangle. i That is, the location of each bright stripe; in this embodiment, D i Approximately the width of each stripe's outline. Since the width of each stripe after segmentation is not necessarily a constant value at different points, it is necessary to divide each stripe according to different distances D. i Perform translations separately. When the width of a stripe remains constant at all points, such as... Figure 6 As shown in the medium gray stripes, the point C where the stripe outline is tangent to the longest side of the smallest bounding rectangle can be used as a reference. i Draw a line segment parallel to the shorter side of the smallest circumscribed rectangle, intersecting the contour at point B. i It intersects the other longer side of the smallest bounding rectangle at point A. i Therefore, it is acceptable. When the width of a stripe is not constant at all points, the translation distance D can be adjusted.i Take the average value of the width of each stripe.

[0064] For simplicity, it is assumed that the width of each stripe is constant D i To solve the problem of 180° ambiguity in the normal direction during translation (a class of stripes translation to another stripe, and the direction of translation is left up or right down), that is Figure 6 The selection of black stripes is shown in the figure. The dark stripe region obtained by score segmentation is set as set m, and the region is translated by distance D i Two regions of the same size can be obtained by translating the region along the inner and outer normal directions, respectively, and are denoted as n and p, forming three regions of the same size, that is, the same number of coordinate point sets; wherein set m corresponds to the dark stripe, set n corresponds to the bright stripe or background region, and set p is determined accordingly; in the original image, the side with a larger inter-class difference is automatically determined as the bright stripe region, and the bright stripe is automatically extracted.

[0065] The inter-class difference is the brightness difference between the bright stripe, the dark stripe, and the other class.

[0066] Embodiment

[0067] This embodiment is aimed at the ocean internal wave occurring at about 02:19:28 on February 23, 2011, which is ALOS satellite data as shown in Figure 1 The image center point has longitude and latitude coordinates of 48°35′W, 29°17′S, and the imaging size is 11200x9600. For convenience of calculation, an image of 1024x1024 in the red box is taken as a sample for analysis, as shown in Figure 2 Synthetic aperture radar ocean internal wave stripe recognition is performed in combination with Gabor transform and K-means clustering.

[0068] The satellite remote sensing experimental data is operated according to the flowchart shown in Figure 3 The specific steps are as follows:

[0069] (1) The original remote sensing data is filtered and preprocessed by ENVI software. All subsequent processing is performed in Matlab software.

[0070] (2) The image obtained in the previous step is used to extract features of the ocean internal wave SAR image by Gabor transform and Gaussian smoothing.

[0071] (3) The K-means clustering algorithm is used to segment the SAR image, so as to separate the bright (dark) stripes of the ocean internal wave from the background, and the center filter segmentation result is compared.

[0072] (4) The result of (3) is compared with the manually extracted image, and the IOU is calculated to obtain the optimal segmentation image.

[0073] (5) Finally, the relative position of the corresponding dark or bright stripe is automatically determined by comparing the inter-class difference, so that the feature extraction of the bright and dark stripes of the ocean internal wave is realized.

[0074] The segmentation result is compared with the ocean internal wave sample image according to the intersection over union (IOU) of the image, the accuracy of segmentation is verified, and the effectiveness and practicability of the method in the bright and dark stripe segmentation of the ocean internal wave SAR image are illustrated, thereby laying a foundation for subsequent ocean internal wave parameter inversion research.

[0075] The electronic device includes a central processing unit (CPU) that can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) or loaded into a random access memory (RAM) from a storage unit. In the RAM, various programs and data required for device operation can also be stored. The CPU, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0076] A plurality of components in the device are connected to the I / O interface, including: an input unit such as a keyboard, a mouse, etc.; an output unit such as various types of displays, a speaker, etc.; a storage unit such as a magnetic disk, an optical disk, etc.; and a communication unit such as a network card, a modem, a wireless communication transceiver, etc. The communication unit allows the device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0077] The processing unit performs various methods and processes described above, such as methods S1-S3. For example, in some embodiments, methods S1-S3 can be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as the storage unit. In some embodiments, part or all of the computer program can be loaded and / or installed on the device via the ROM and / or the communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of methods S1-S3 described above can be performed. Alternatively, in other embodiments, the CPU can be configured to perform methods S1-S3 by any other appropriate means (e.g., by means of firmware).

[0078] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that can be used include Field-programmable Gate Arrays (FPGAs), Application-specific Integrated Circuits (ASICs), Application-specific Standard Products (ASSPs), System-on-a-chip systems (SOCs), Complex Programmable Logic Devices (CPLDs), etc.

[0079] Program code for carrying out methods of the present application can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces a means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. The program code can be retrieved from a machine-readable medium, a machine-readable storage medium, or a machine-readable storage device. The machine-readable medium can be embodied in an article of manufacture that can be used with an apparatus, such as a machine, an apparatus, or a manufacturing machine. The machine-readable medium can be a machine-readable storage medium or a machine-readable storage device incorporated with a corresponding service or functionality.

[0080] In the context of the present application, a machine-readable medium can be a tangible medium that can contain or store program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable storage medium can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium would include one or more lines of electrical connections, portable computer disks, hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0081] The above description is only specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for segmenting oceanic internal wave ribbons in SAR imagery, the method comprising: The method comprises the following steps: Step S1, acquiring an ocean internal wave SAR image, and performing feature extraction on the ocean internal wave SAR image by using Gabor transformation to obtain a feature image; Step S2, performing stripe segmentation on the feature image by using a K-means clustering algorithm to separate bright stripes from background in the ocean internal wave and automatically determine a region where dark stripes are located according to an inter-class difference, or to separate dark stripes from background in the ocean internal wave and automatically determine a region where bright stripes are located according to the inter-class difference; Step S3, based on the region where the dark stripes are located, constructing a minimum circumscribed rectangle of a bright stripe contour, and translating a set distance along a normal direction of a long side of the minimum circumscribed rectangle to obtain a position where the dark stripes are located, or based on the region where the bright stripes are located, constructing a minimum circumscribed rectangle of a dark stripe contour, and translating a set distance along a normal direction of a long side of the minimum circumscribed rectangle to obtain a position where the bright stripes are located; In the step S1, the Gabor transformation is used to perform feature extraction on the ocean internal wave SAR image to obtain the feature image, and the feature image is obtained by the following steps: based on a corresponding relationship between a Gabor kernel function and a filter image, a Gabor kernel function representing different wavelengths and directions is constructed, and is respectively convolved with the ocean internal wave SAR image to obtain multi-dimensional filter images responding to different features under different Gabor kernel functions, and then the multi-dimensional filter images are synthesized into a one-dimensional image through Gaussian smoothing, and a feature image used for classification is extracted.

2. The method of claim 1, wherein, The Gabor transformation is an even-symmetry filter taking only a real part of a Gabor filter, and a real part expression of the corresponding Gabor filter is: where x, y are pixel position coordinates, is the new coordinate generated by Gabor kernel function rotation, is the wavelength of the cosine function; is the normal direction of the parallel strip, i.e. the texture direction, which is selected at a set interval between 0°-180°; is the phase parameter of the cosine function, value is 0°; is the standard deviation of the Gaussian function; is the space aspect ratio of the Gabor function.

3. The method of claim 2, wherein the method further comprises: The cosine function wavelength The correspondence with spatial frequency is: where the spatial frequency is normalized to be between -0.5 and 0.5, for spatial frequencies between 0 and 0.25, for spatial frequencies between 0.25 and 0.5; for image size; where where b is the half response spatial frequency bandwidth.

4. The method of claim 2, wherein the method further comprises: The texture direction in the step S1 The selection is made at intervals of 30° between 0° and 180°.

5. The method of claim 1, wherein the method further comprises: A Gaussian function expression used in the Gaussian smoothing is: In the formula, is the standard deviation of the Gaussian window, and x, y are the pixel position coordinates.

6. The method of claim 1, wherein, In the step S2, the K-means clustering algorithm is used to perform stripe segmentation on the feature image, and the stripe segmentation is performed by the following steps: 1) initializing a number of clusters K, randomly creating an initial division, and giving an initial center of clustering; 2) improving the division effect by constantly moving the center of clustering through an iterative method until a criterion function converges, wherein, an expression of the criterion function is: wherein is the sum of squared errors, is the number of classes, is the current class value, is the class average.

7. The method of claim 1, wherein the method further comprises: In the step S3, based on the region where the bright stripes are located, the minimum circumscribed rectangle of the dark stripe contour is constructed, and the set distance is translated along the normal direction of the long side of the minimum circumscribed rectangle to obtain the position where the bright stripes are located, and the stripe segmentation is performed by the following steps: Based on the image characteristics of oceanic internal wave, the minimum circumscribed rectangle of each dark fringe profile is used to translate a certain distance along the normal direction of the long side of the rectangle The position of each bright fringe is obtained; the dark fringe region is divided into a set m, and the region is translated a distance along the inner and outer normal directions Two regions of the same size are obtained, which are denoted as n and p, forming three regions of the same size, i.e., the same number of coordinate point sets; among them, set m corresponds to the dark fringe, set n corresponds to the bright fringe or background region, and set p is determined accordingly; in the original image, the side with a larger inter-class difference is automatically determined as the bright fringe region, and the bright fringe is automatically extracted. 8.An electronic device comprising a memory and a processor, the memory having stored thereon a computer program, characterized in that, The processor executes the program to implement the method according to any one of claims 1-7.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the method according to any one of claims 1-7.

Images