Ship Detection Method in Synthetic Aperture Radar Images

Abstract

The disclosed embodiment relates to a method for detecting ships in synthetic aperture radar images, the method comprising: dividing the original SAR image into P sub-blocks, performing singular value processing on each sub-block, and obtaining the singular value feature combination of each sub-block in the original SAR image; performing cluster analysis on the singular value feature combination of all sub-blocks, and obtaining the mask image of the candidate area of ​​the ship; performing superpixel segmentation on the original SAR image and binarizing the edges of the segmented sub-images to obtain the segmented SAR image, wherein the boundary area of ​​each sub-image in the segmented SAR image is the binarized information; performing intersection fusion processing on the mask image of the candidate area of ​​the ship and the segmented SAR image, determining the closed area of ​​the intersection fusion processing, and using the closed area as the detected ship. The method can effectively solve the problem of interference of sea clutter on ship detection, and realize accurate detection of ships on the sea surface.

Description

Technical Field

[0001] The present application belongs to the field of image processing technology, and in particular relates to a method for detecting ships in synthetic aperture radar (SAR) images. Background Art

[0002] The sea surface ship detection technology is of great significance to the fields of port management, environmental monitoring, and maritime law enforcement. Synthetic aperture radar (SAR) is an active microwave imaging system that works around the clock and is not affected by weather conditions. It is widely used in ship detection. Ships are mostly made of metal materials. Therefore, in the ship area, the SAR system will receive a strong echo signal, while the sea surface has a strong scattering effect on electromagnetic signals, so the received echo signal is weaker. However, due to the presence of sea clutter, the sea surface area will also produce high-intensity signals, and a lot of speckle noise will appear, which will greatly interfere with ship detection, especially in poor sea conditions, the interference will be stronger.

[0003] Traditional ship detection methods mostly rely on SAR intensity information. However, when the intensity of sea surface noise is close to the ship signal intensity, a large number of false alarms will be generated, seriously affecting the effect of ship detection. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present application provides a method for detecting ships in SAR images, which can effectively solve the problem of interference of sea clutter on ship detection and realize accurate detection of ships on the sea surface.

[0006] (II) Technical solution

[0007] In order to achieve the above objectives, this application adopts the following technical solutions:

[0008] In a first aspect, the present application provides a method for detecting a ship in a SAR image, the method comprising:

[0009] A10, dividing the original synthetic aperture radar SAR image into P sub-blocks, performing singular value processing on each sub-block, and obtaining a singular value feature combination of each sub-block in the original SAR image;

[0010] A20, performing cluster analysis on the singular value feature combinations of all sub-blocks to obtain a mask image of the candidate ship area;

[0011] A30, performing superpixel segmentation on the original SAR image and performing binarization processing on the edges of the segmented sub-images to obtain a segmented SAR image, wherein the boundary area of ​​each sub-image in the segmented SAR image is binarization information;

[0012] A40, performing intersection fusion processing on the mask image of the candidate ship area and the segmented SAR image, determining a closed area of ​​the intersection fusion processing, and using the closed area as the detected ship.

[0013] Optionally, the A10 includes:

[0014] A11, dividing the original SAR image into P sub-blocks in a manner of N*N size and a step size of 1;

[0015] A12. For each sub-block G (i,j) , the center pixel coordinates are (i, j) and the window size is N;

[0016] U,S,V T =SVD(G (i,j) );

[0017]

[0018] Where SVD(*) represents the singular value decomposition operation, S is a singular value matrix or vector, containing all singular values ​​s1,s2,...,s N ,

[0019] U, V T G (i,j) G (i,j)T and G (i,j)T G (i,j) The corresponding feature vector set;

[0020] Get U, S, V of each sub-block T ;

[0021] A13. For each sub-block S, three singular value features are obtained for measuring the local density information of the image.

[0022] A14, combining the three singular value features of each sub-block with the intensity information of the sub-block in the original SAR image to obtain a singular value feature combination of each sub-block. It can be understood that the central pixel value of the image sub-block is combined with the three singular value features corresponding to the sub-block.

[0023] Optionally, the A13 includes:

[0024] For a sub-block, F1, F2, and F3 are three singular value features of the sub-block;

[0025] F1 = s1 - s2;

[0026] F2=s2 / s1;

[0027] F3=∑ k s k ;

[0028] k=1,2,…,N;

[0029] Among them, k represents the index of the singular value in the singular value matrix S, and the difference between s1 and s2 represents the difference in energy density between the ship and the sea clutter.

[0030] Optionally, the A14 includes:

[0031] For an original SAR image with M pixels, the singular value feature combination F of each sub-block is expressed as:

[0032]

[0033] Among them, I m represents the pixel intensity of the mth pixel in the original SAR image, and I represents each pixel in the original SAR image;

[0034] F1 m 、F2 m 、F3 m They respectively represent the three singular value features corresponding to a sub-block when the m-th pixel in the original SAR image is taken as the central pixel of the sub-block.

[0035] Optionally, the A20 includes:

[0036] The K-means clustering method is used to perform cluster analysis on the singular value feature combinations of all sub-blocks to obtain the mask image of the ship candidate area (i.e., the binary image of the ship-sea surface).

[0037] Optionally, the A30 includes:

[0038] A31, performing superpixel segmentation on the original SAR image to obtain Q segmented sub-images;

[0039] A32, performing binarization processing on the edge area of ​​each sub-image;

[0040] A33. All sub-images after binarization are combined into a segmented SAR image.

[0041] Optionally, the A40 includes:

[0042] The mask image and the segmented SAR image are subjected to intersection fusion processing according to the following formula;

[0043]

[0044] I final =∪ C SP C

[0045] Among them, SP C represents the Cth superpixel in the image, (i,j) represents the pixel coordinates in the image, and I ship According to the mask image, I final Indicates a closed region.

[0046] In a second aspect, the present application provides an electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method for detecting ships in SAR images as described in any one of the first aspects above.

[0047] In a third aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method for detecting ships in SAR images as described in any one of the first aspects above are implemented.

[0048] (III) Beneficial effects

[0049] The technical solution provided by this application may have the following beneficial effects:

[0050] The present invention proposes to detect energy density by adopting a method of local singular value analysis of an image, and suppress the intensity of sea clutter by extracting effective singular value features, thereby improving the signal-to-noise ratio of a ship image.

[0051] Compared with the traditional ship detection method based on pixel intensity, this method fully considers the differences in spatial distribution characteristics of sea clutter and ships on SAR images, and uses local neighborhood features to effectively solve the problems of high false alarm rate and difficulty in accurately locating the ship position in the ship detection process under obvious noise conditions.

[0052] The present invention adopts the K-means unsupervised learning method, which can adaptively realize the classification process without manually marking samples. It is not affected by sea conditions, electromagnetic wave polarization mode, sensor resolution, etc., and is convenient and efficient.

[0053] The present invention adopts the superpixel segmentation method to obtain the precise contour information of the ship, and fuses it with the candidate area to obtain a high-confidence ship detection result, which effectively compensates for the problem of ship edge expansion and blurring during the singular value analysis process.

[0054] Based on actual operation, it can be known that the value of the first singular value of the ship area is very large compared to the value of the second singular value, and its influence on the entire detection and identification is relatively large. For this reason, the present invention adopts the method of taking the difference and quotient of the first singular value (such as s1) and the second singular value (s2) as two effective features for distinguishing the sea surface from the ship. In particular, since the energy density of the ship area is concentrated and the sum of the singular values ​​is greater than that of the sea surface area, this information is also used as a feature for subsequent processing in this embodiment, thereby effectively improving the accuracy of ship detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The present application is described with the aid of the following drawings:

[0056] Figure 1 A schematic diagram of the process of detecting a ship in a SAR image proposed by the present invention;

[0057] Figure 2 A schematic diagram of an original SAR image provided by an embodiment of the present invention;

[0058] Figure 3 A schematic diagram of a singular value feature image provided by an embodiment of the present invention;

[0059] Figure 4 A schematic diagram of a mask image of a candidate ship region provided by an embodiment of the present invention;

[0060] Figure 5 A schematic diagram of a superpixel segmentation result provided by an embodiment of the present invention;

[0061] Figure 6 Schematic diagram of the final ship detection result of an embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below in conjunction with the accompanying drawings through specific implementation methods. It is understood that the specific embodiments described below are only used to explain the relevant inventions, rather than to limit the invention. It should also be noted that the embodiments and features in the embodiments of this application can be combined with each other in the absence of conflict; for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0063] At present, the energy density of the ship area image is relatively concentrated, while the signal energy distribution of the sea clutter is relatively sparse. Therefore, the energy density information can be used as a feature to effectively distinguish between ships and sea clutter. Local singular value analysis can effectively detect the energy density in the image, and superpixel segmentation can effectively obtain the outline of the ship. The combination of the two can effectively weaken the impact of sea clutter on ship detection and obtain accurate ship detection results. To this end, this embodiment provides a method for detecting ships in SAR images based on the above concept, which uses a method of local singular value analysis of the image to detect energy density, and suppresses the intensity of sea clutter by extracting effective singular value features, thereby improving the signal-to-noise ratio of the ship image.

[0064] Embodiment 1

[0065] Figure 1 FIG. 1 is a flow chart of a method for detecting a ship in a SAR image in an embodiment of the present application. The method of the present embodiment can be executed by any computing device, and the computing device can be implemented in the form of software and / or hardware, such as Figure 1 As shown, the method comprises the following steps:

[0066] A10, dividing the original synthetic aperture radar SAR image into P sub-blocks, performing singular value processing on each sub-block, and obtaining a singular value feature combination of each sub-block in the original SAR image;

[0067] A20, performing cluster analysis on the singular value feature combinations of all sub-blocks to obtain a mask image of the candidate ship area;

[0068] For example, the K-means clustering method is used to perform cluster analysis on the singular value feature combinations of all sub-blocks to obtain the mask image of the candidate ship area.

[0069] A30, performing superpixel segmentation on the original SAR image and performing binarization processing on the edges of the segmented sub-images to obtain a segmented SAR image, wherein the boundary area of ​​each sub-image in the segmented SAR image is binarization information;

[0070] A40, performing intersection fusion processing on the mask image of the candidate ship area and the segmented SAR image, determining a closed area of ​​the intersection fusion processing, and using the closed area as the detected ship.

[0071] Compared with the traditional ship detection method based on pixel intensity, the method of this embodiment fully considers the differences in spatial distribution characteristics of sea clutter and ships on SAR images, and uses local neighborhood features to effectively solve the problems of high false alarm rate and difficulty in accurately locating the ship position in the ship detection process under obvious noise conditions.

[0072] Combination Figures 2 to 6The method of the embodiment of the present invention is described in detail.

[0073] The method of this embodiment includes the following steps:

[0074] A11, dividing the original SAR image into P sub-blocks in a manner of N*N size and a step size of 1;

[0075] A12. For each sub-block G (i,j) , the center pixel coordinates are (i, j), the window size is N (such as N = 1); Figure 2 shown.

[0076] U,S,V T =SVD(G (i,j) );

[0077]

[0078] Where SVD(*) represents the singular value decomposition operation, S is a singular value matrix or vector, containing all singular values ​​s1,s2,...,s N ,s k (k=1,2,...,N) are singular values.

[0079] U, V T G (i,j) G (i,j)T and G (i,j)T G (i,j) The corresponding feature vector set;

[0080] Get U, S, V of each sub-block T ;

[0081] A13. For each sub-block S, three singular value features are obtained for measuring the local density information of the image.

[0082] For a sub-block, F1, F2, and F3 are three singular value features of the sub-block, which are local singular value features that can effectively distinguish between ships and sea clutter;

[0083] F1 = s1 - s2;

[0084] F2=s2 / s1;

[0085] F3=∑ k s k ;

[0086] k=1,2,…,N;

[0087] Among them, k represents the index of the singular value in the singular value matrix S, the difference between s1 and s2 represents the difference in energy density between the ship and the sea clutter, and at the same time, the sum of the singular values ​​(F3) also has the ability to distinguish, so in this embodiment, the user processes the three information to achieve subsequent further precision detection.

[0088] A14, combining the three singular value features of each sub-block with the intensity information of the sub-block in the original SAR image to obtain a singular value feature combination of each sub-block. The specific operation is to combine the central pixel value of the image sub-block with the three singular value features corresponding to the sub-block. In other words, the intensity information of the sub-block in the original SAR image is the pixel value of the central pixel of the area corresponding to the sub-block.

[0089] For an original SAR image with M pixels, the singular value feature combination F of each sub-block is expressed as:

[0090]

[0091] Among them, I m represents the pixel intensity of the mth pixel in the original SAR image, and M is the total number of pixels in the image. Figure 3 shown.

[0092] F1 m 、F2 m 、F3 m They respectively represent the three singular value features corresponding to a sub-block when the m-th pixel in the original SAR image is taken as the central pixel of the sub-block.

[0093] A20. Use K-means clustering method to perform cluster analysis on the singular value feature combinations of all sub-blocks to obtain the mask image of the candidate ship area, that is, the ship-sea surface binary image, such as Figure 4 shown.

[0094] A31, performing superpixel segmentation on the original SAR image to obtain Q segmented sub-images;

[0095] A32, performing binarization processing on the edge area of ​​each sub-image;

[0096] A33. All sub-images after binarization are combined into a segmented SAR image.

[0097] The original SAR image is segmented by SLIC superpixel to obtain the edge information of the superpixel sub-image, which contains the high-precision ship boundary. In this embodiment, the number of superpixel seed points is set to 100. The schematic diagram of the superpixel segmentation result is shown in FIG. Figure 5 As shown in (a), the superpixel boundary binarization image is as follows Figure 5 (b) as shown.

[0098] It should be noted that the edge information is the boundary of the superpixel block. Superpixel segmentation can be a local clustering algorithm. First, a considerable number of initialization seed points are evenly set in the image. Pixels with similar features around the seed points are divided into a class through unsupervised clustering methods. This type of pixel cluster is called a "superpixel". Therefore, the boundary of the superpixel often corresponds to the edge of the object and the gradient change information in the image, that is, the edge contour of the ship is contained in Figure 5 However, superpixel is an edge segmentation method, which cannot perform target detection tasks when used alone. Therefore, it is necessary to combine other steps of this embodiment to obtain a mask image of the ship area (i.e., a rough detection result). In this embodiment, the superpixel method is used to obtain a better target detection result.

[0099] A40, performing intersection fusion processing on the mask image and the segmented SAR image according to the following formula;

[0100]

[0101] I final =∪ C SP C

[0102] Among them, SP C represents the Cth superpixel in the image (i.e., a certain superpixel), (i,j) represents the pixel coordinates in the image, I ship According to the mask image, I final Represents a closed area. The fusion result is as follows Figure 6 Finally, through the solution of the present invention, accurate recognition results of ships on the sea surface are obtained, and the interference of sea surface noise is completely eliminated.

[0103] Aiming at the problem that sea clutter in SAR images can easily interfere with ship detection, the present invention proposes a synthetic aperture radar sea surface ship detection method combining local singular value analysis and superpixel segmentation. The singular value analysis method is used to extract the local spatial distribution characteristics of the image, and K-means clustering is used for binary classification. At the same time, superpixel segmentation is used to obtain the ship contour. By fusing the above two results, high-precision sea surface ship detection can be achieved.

[0104] Embodiment 3

[0105] In a third aspect, the present application provides an electronic device through Embodiment 3, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the method for detecting ships in SAR images as described in any one of the above embodiments are implemented.

[0106] The method disclosed in the above embodiment of the present invention can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a ready-made programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The methods, steps, and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present invention can be directly embodied as being executed by a hardware decoding processor, or executed by a combination of hardware and software units in a decoding processor. The software unit may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor 101 reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0107] In addition, in combination with the method for detecting ships in SAR images in the above embodiments, an embodiment of the present invention may provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, any method for detecting ships in SAR images in the above method embodiments is implemented.

[0108] It should be noted that in the claims, any figure marks between brackets should not be understood as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "one" or "an" preceding a component does not exclude the presence of multiple such components. In addition, it should be noted that in the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" and the like refers to the specific features, structures, materials or characteristics described in conjunction with the embodiment or example included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0109] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments after knowing the basic creative concept. Therefore, the claims should be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0110] Obviously, those skilled in the art can make various modifications and variations 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 should also include these modifications and variations.

Claims

1. A method for detecting ships in synthetic aperture radar images, characterized in that: The method includes: A10, dividing the original synthetic aperture radar SAR image into P sub-blocks, performing singular value processing on each sub-block, and obtaining a singular value feature combination of each sub-block in the original SAR image; A20, performing cluster analysis on the singular value feature combinations of all sub-blocks to obtain a mask image of the candidate ship area; A30, performing superpixel segmentation on the original SAR image and performing binarization processing on the edges of the segmented sub-images to obtain a segmented SAR image, wherein the boundary area of ​​each sub-image in the segmented SAR image is binarization information; A40, performing intersection fusion processing on the mask image of the candidate ship area and the segmented SAR image, determining a closed area of ​​the intersection fusion processing, and using the closed area as the detected ship; The A10 includes: A11, dividing the original SAR image into P sub-blocks in a manner of N*N size and a step size of 1; A12. For each sub-block G (i,j) , the center pixel coordinates are (i, j) and the window size is N; U,S,V T =SVD(G (i,j) ); Get U, S, V of each sub-block T ; where SVD(*) represents the singular value decomposition operation, S is a singular value matrix or vector, containing all singular values ​​s1,s2,...,s N , U, V T G (i,j) G (i,j)T and G (i,j)T G (i,j) The corresponding feature vector set; A13. For each sub-block S, three singular value features are obtained for measuring the local density information of the image. A14, combining the three singular value features of each sub-block and the intensity information of the sub-block in the original SAR image to obtain a singular value feature combination of each sub-block; A13 includes: for a sub-block, F1, F2, and F3 are three singular value features of the sub-block; Among them, k represents the index of the singular value in the singular value matrix S, and the difference between s1 and s2 represents the difference in energy density between the ship and the sea clutter.

2. The method according to claim 1, characterized in that The A14 includes: For an original SAR image with M pixels, the singular value feature combination F of each sub-block is expressed as: Among them, I m represents the pixel intensity of the mth pixel in the original SAR image, and I represents each pixel in the original SAR image; F1 m 、F2 m 、F3 m They respectively represent the three singular value features corresponding to a sub-block when the m-th pixel in the original SAR image is taken as the central pixel of the sub-block.

3. The method according to claim 2, characterized in that The A20 includes: The K-means clustering method is used to perform cluster analysis on the singular value feature combinations of all sub-blocks to obtain the mask image of the candidate ship area.

4. The method according to claim 2, characterized in that: The A30 includes: A31, performing superpixel segmentation on the original SAR image to obtain Q segmented sub-images; A32, performing binarization processing on the edge area of ​​each sub-image; A33. All sub-images after binarization are combined into a segmented SAR image.

5. The method according to any one of claims 1 to 4, characterized in that: The A40 includes: The mask image and the segmented SAR image are subjected to intersection fusion processing according to the following formula; Among them, SP C represents the Cth superpixel in the image, (i,j) represents the pixel coordinates in the image, and I ship According to the mask image, I final Indicates a closed region.

6. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method for detecting a ship in a synthetic aperture radar image as described in any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for detecting a ship in a synthetic aperture radar image as described in any one of claims 1 to 5 are implemented.

Images