Method, device and equipment for separating ship and chaff in SAR image

Abstract

The specification provides a ship and chaff segmentation method, device and equipment in a SAR image. The method comprises: constructing a coherence matrix of each pixel according to the full polarization data of a SAR radar, and decomposing the coherence matrix into the following scattering mechanisms: surface scattering, secondary scattering and volume scattering, and then generating a pseudo-color image based on the power proportion of each scattering mechanism; next, by applying the foreground mask of the SAR image to the pseudo-color image, a to-be-classified image is obtained; finally, the pixels in the to-be-classified image are classified based on a preset scattering characteristic difference and marked as ship category pixels and chaff category pixels. Through the synergistic effect of polarization scattering mechanism decomposition and mask noise reduction, high-precision differentiation of ship and chaff interference is realized.

Description

Technical Field

[0001] This specification relates to one or more embodiments in the field of radar signal processing, and more particularly to a method, apparatus and device for segmenting ships and chaff in SAR images. Background Technology

[0002] Compared to other guidance technologies, SAR seekers, which use Synthetic Aperture Radar (SAR) technology as their core sensor, can effectively improve the target detection accuracy and recognition probability of the seeker, and have significant advantages in many aspects: all-weather and all-time imaging capability, long detection range, and provision of richer target features.

[0003] Chalcoaching is a crucial jamming method against anti-ship missiles. Centroid jamming, in particular, is a jamming pattern used to counter radar seekers during the tracking phase. Its purpose is to make the radar seeker track the target ship and the energy centroid of the chaff cloud. As the chaff cloud drifts with the wind and the target ship maneuvers, the radar seeker's tracking point gradually falls onto the chaff cloud and moves away from the target ship. Ultimately, the target ship moves away from the radar seeker's tracking beam, thus avoiding attack. Figure 1 As shown.

[0004] In the initial stage of a ship deploying chaff for centroid-based jamming, when the ship and chaff are not completely separated in SAR images, existing detection methods, such as Constant False Alarm Rate (CFAR) detection, may mistake the chaff and the ship for the same target. As a result, missiles may fail to hit the ship during an attack. Therefore, it is necessary to identify and segment the detected area, which is crucial for the accurate hit of anti-ship missiles. Summary of the Invention

[0005] In view of this, one or more embodiments of this specification provide a method, apparatus, device and storage medium for segmenting ships and chaff in SAR images, so as to achieve accurate segmentation of ships and chaff.

[0006] To achieve the above objectives, one or more embodiments of this specification provide the following technical solutions:

[0007] According to a first aspect of one or more embodiments of this specification, a method for segmenting ships and chaff in SAR images is proposed, comprising:

[0008] Acquire full polarization data from SAR radar;

[0009] The coherence matrix of each pixel is constructed based on the fully polarized data, and the coherence matrix is ​​decomposed into the following scattering mechanisms: surface scattering, secondary scattering, and volume scattering;

[0010] Pseudo-color images are generated based on the power proportion of each scattering mechanism;

[0011] Obtain the foreground mask in the SAR image corresponding to the fully polarimetric data, and apply the foreground mask to the pseudo-color image to obtain the image to be classified;

[0012] The pixels in the image to be classified are classified based on the differences in preset scattering features and labeled as ship category pixels and foil category pixels to obtain the segmentation result.

[0013] In some embodiments, the coherence matrix is ​​decomposed into the following scattering mechanisms: surface scattering, secondary scattering, and volume scattering, including:

[0014] The coherence matrix is ​​equivalent to a weighted combination of planar scattering power, secondary scattering power, and volume scattering power.

[0015] The total power of a SAR image is equal to the sum of the plane scattering power, the secondary scattering power, and the volume scattering power, along with the weight values ​​corresponding to the three types of scattering power, and each scattering power is solved.

[0016] In some embodiments, obtaining the foreground mask in the SAR image corresponding to the fully polarimetric data includes:

[0017] Obtain the total power map of the SAR image;

[0018] A threshold is set based on the target features, and the total power map is binarized using the threshold.

[0019] The isolated points in the binarization result are removed to obtain the foreground mask.

[0020] In some embodiments, the step of setting a threshold based on target features and binarizing the total power map using the threshold includes: performing constant false alarm rate detection on the total power map to perform binarization;

[0021] The process of removing isolated points from the binarization result to obtain the foreground mask includes:

[0022] The foreground mask is obtained by removing isolated points from the binarization result through morphological opening operations.

[0023] In some embodiments, classifying the pixels in the image to be classified based on preset scattering feature differences and labeling them as ship category pixels and chaff category pixels includes:

[0024] The pixels in the image to be classified are classified using a support vector machine, wherein the support vector machine is trained using the power proportion values ​​of the scattering mechanisms of multiple ships and multiple foil strips.

[0025] In some embodiments, the method further includes:

[0026] The image to be filtered is obtained based on the label value of each pixel in the segmentation result, wherein the label value of the background pixel is zero and the label value of the foreground pixel is non-zero.

[0027] For each non-zero pixel, select a neighborhood window of a fixed size;

[0028] Within the neighborhood window, the current pixel is updated using the most non-zero label value in the connected neighborhood of the current non-zero pixel.

[0029] According to a second aspect of one or more embodiments of this specification, a device for segmenting ships and chaff in SAR images is provided, comprising:

[0030] Acquisition unit, used to acquire full polarization data of SAR radar;

[0031] The decomposition unit is used to construct the coherence matrix of each pixel based on the fully polarized data, and decompose the coherence matrix into the following scattering mechanisms: surface scattering, secondary scattering and volume scattering;

[0032] The generation unit is used to generate pseudo-color images based on the power proportion of each scattering mechanism;

[0033] A masking unit is used to acquire the foreground mask in the SAR image corresponding to the fully polarimetric data, and apply the foreground mask to the pseudo-color image to obtain the image to be classified.

[0034] The segmentation unit is used to classify the pixels in the image to be classified based on the differences in preset scattering features, and label them as ship category pixels and foil category pixels to obtain the segmentation result.

[0035] In some embodiments, the decomposition unit is used for:

[0036] The coherence matrix is ​​equivalent to a weighted combination of planar scattering power, secondary scattering power, and volume scattering power.

[0037] The total power of a SAR image is equal to the sum of the plane scattering power, the secondary scattering power, and the volume scattering power, along with the weight values ​​corresponding to the three types of scattering power, and each scattering power is solved.

[0038] In some embodiments, the mask unit is used for:

[0039] Obtain the total power map of the SAR image;

[0040] A threshold is set based on the target features, and the total power map is binarized using the threshold.

[0041] The isolated points in the binarization result are removed to obtain the foreground mask.

[0042] In some embodiments, when the mask unit is used to set a threshold based on target features and to binarize the total power map using the threshold, it is specifically used to: perform constant false alarm rate detection on the total power map for binarization;

[0043] The mask unit is used to obtain the foreground mask by removing isolated points from the binarization result, specifically as follows:

[0044] The foreground mask is obtained by removing isolated points from the binarization result through morphological opening operations.

[0045] In some embodiments, classifying the pixels in the image to be classified based on preset scattering feature differences and labeling them as ship category pixels and chaff category pixels includes:

[0046] The pixels in the image to be classified are classified using a support vector machine, wherein the support vector machine is trained using the power proportion values ​​of the scattering mechanisms of multiple ships and multiple foil strips.

[0047] In some embodiments, the device further includes a filtering unit for:

[0048] The image to be filtered is obtained based on the label value of each pixel in the segmentation result, wherein the label value of the background pixel is zero and the label value of the foreground pixel is non-zero.

[0049] For each non-zero pixel, select a neighborhood window of a fixed size;

[0050] Within the neighborhood window, the current pixel is updated using the most non-zero label value in the connected neighborhood of the current non-zero pixel.

[0051] According to a third aspect of one or more embodiments of this specification, an electronic device is provided, comprising:

[0052] processor;

[0053] Memory used to store processor-executable instructions;

[0054] The processor implements the steps of the method proposed in the above embodiments by running the executable instructions.

[0055] According to a fourth aspect of one or more embodiments of this specification, a computer-readable storage medium is provided that stores computer instructions thereon, which, when executed by a processor, implement the steps of the method proposed in the above embodiments.

[0056] According to a fifth aspect of one or more embodiments of this specification, a computer program product is provided, comprising a computer program / instructions that, when executed by a processor, implement the steps of the method proposed in the above embodiments.

[0057] The method for segmenting ships and chaff in SAR images proposed in this disclosure constructs a coherence matrix for each pixel based on the full polarization data of the SAR radar, and decomposes the coherence matrix into the following scattering mechanisms: surface scattering, secondary scattering, and volume scattering. Then, a pseudo-color image is generated based on the power proportion of each scattering mechanism. Next, a foreground mask of the SAR image is applied to the pseudo-color image to obtain the image to be classified. Finally, the pixels in the image to be classified are classified based on preset scattering feature differences and labeled as ship category pixels and chaff category pixels. This scheme achieves high-precision differentiation between ship and chaff interference through the synergistic effect of polarization scattering mechanism decomposition and mask noise reduction. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the centroid interference of the foil strip.

[0059] Figure 2 This is a flowchart of a method for segmenting ships and chaff in a SAR image, provided as an exemplary embodiment.

[0060] Figure 3 This is an exemplary pseudo-color image obtained from a fully polarized exploration of an ocean area.

[0061] Figure 4A This is an example HH channel map in a fully polarimetric SAR image.

[0062] Figure 4B This is an example HV channel map in a fully polarimetric SAR image.

[0063] Figure 4C This is an example VV channel map in a fully polarimetric SAR image.

[0064] Figure 5 This is a span total power plot of an exemplary fully polarimetric SAR image.

[0065] Figure 6 This is an example foreground mask.

[0066] Figure 7 This is an example of the segmentation result of a ship and chaff.

[0067] Figure 8 This is an example of a window filtering diagram.

[0068] Figure 9 This is another exemplary window filtering diagram.

[0069] Figure 10 This is a schematic diagram of an example filtering result.

[0070] Figure 11 This is a block diagram of a SAR image segmentation device for ships and chaff in an exemplary embodiment.

[0071] Figure 12 This is a schematic diagram of the structure of a device provided in an exemplary embodiment. Detailed Implementation

[0072] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0073] It should be noted that the steps of the corresponding methods are not necessarily performed in the order shown and described in this specification in other embodiments. In some other embodiments, the methods may include more or fewer steps than described in this specification. Furthermore, a single step described in this specification may be broken down into multiple steps in other embodiments; and multiple steps described in this specification may be combined into a single step in other embodiments.

[0074] Figure 2 A flowchart illustrating a method for segmenting ships and chaff in SAR images according to an embodiment of this disclosure is shown. Figure 2 As shown, the method includes:

[0075] Step 201: Obtain the full polarization data of the SAR radar.

[0076] Fully polarimetric SAR acquires the fully polarimetric scattering information of a target by transmitting and receiving electromagnetic waves with different polarizations. Its core data is the scattering matrix, represented as:

[0077]

[0078] Among them, S HH This indicates that the radar emits horizontally polarized (H) electromagnetic waves and receives the horizontally polarized (H) component reflected from the target; S HV This indicates that the radar emits horizontally polarized (H) electromagnetic waves but receives the vertically polarized (V) components reflected from the target; S VHThis indicates that the radar emits vertically polarized (V) electromagnetic waves and receives the horizontally polarized (H) component reflected from the target; S VV This indicates that the radar emits vertically polarized (V) electromagnetic waves and receives the vertically polarized (V) components reflected from the target. For radars where the transmitting and receiving are at the same location, the polarization scattering matrix satisfies the reciprocity condition: S HV =S VH That is, the target responds the same way to horizontal transmit-vertical receive (HV) and vertical transmit-horizontal receive (VH).

[0079] Step 201: Construct a coherence matrix based on the fully polarized data, and decompose the coherence matrix into the following scattering mechanisms: surface scattering, secondary scattering, and volume scattering.

[0080] The formula for the three-dimensional polarimetric coherence matrix for each pixel in the fully polarimetric data is as follows:

[0081]

[0082] Where k is the polarization scattering vector, as shown in the following formula:

[0083]

[0084] Substituting k into formula (2), we obtain the S-form expression of the coherence matrix as follows:

[0085]

[0086] The coherence matrix (4) is incoherently decomposed into the sum of surface scattering, secondary scattering, and volume scattering components:

[0087] T = P s T s +P d T d +P v T v (5)

[0088] Among them, P s P d P v T represents the power of plane scattering, secondary scattering, and volume scattering, respectively. All three are non-negative real numbers, and their sum is equal to the total polarization power span. s T d T v These are coherent matrix templates representing surface scattering, secondary scattering, and volume scattering, respectively. These templates are predefined based on physical scattering models and are used to decompose the scattering mechanism of a target into a weighted combination of the three types.

[0089] This power conservation characteristic can be described as follows:

[0090] Span = P s +P d +P v =T 11 +T 22 +T 33 (6)

[0091] Among them, T 11 T represents the surface scattering power. 22 T represents the secondary scattering power. 33 This represents the volume scattering power.

[0092] In some embodiments, the plane scattering component can be modeled using a first-order Bragg scattering model, with the corresponding T... s It can be represented as follows:

[0093]

[0094] The secondary scattering component can be modeled using a dihedral reflection model, and the corresponding T d It can be represented as follows:

[0095]

[0096] In formulas (7) and (8), α and β are both complex numbers with absolute values ​​less than 1. α can be determined by material properties (dielectric constant), geometric parameters (angle, height difference), and phase effects, while β can be determined by the surface dielectric constant, incident angle, and Bragg scattering physics model, or by data decomposition and inversion.

[0097] Volume scattering corresponds to scatterers with random orientations. These scatterers are spatially chaotic, with random orientation angles (uniformly distributed from 0 to 2π), and lack phase coherence (incoherent superposition). Therefore, the model of volume scattering is an incoherent superposition of dipoles with orientation angles uniformly distributed between 0 and 2π. Due to the presence of numerous dipoles and the random distribution of orientation angles, the target is highly chaotic. The corresponding T... v It can be represented as follows:

[0098]

[0099] Step 203: Generate a pseudo-color image based on the power proportion of each scattering mechanism.

[0100] By scattering P from the surface s Secondary scattering P d Volume scattering P v The power percentage is mapped to the RGB color channels to obtain a pseudo-color image. This image can represent the dominant regions of different scattering mechanisms.

[0101] Figure 3The image shown is an exemplary pseudo-color image obtained from a fully polarimetric detection of an ocean area. The red channel represents secondary scattering, the green channel represents volume scattering, and the blue channel represents surface scattering. The values ​​of each channel were normalized during the synthesis of the pseudo-color image. It can be observed that sea surface scattering is mainly surface scattering; chaff scattering is mainly volume scattering (green area), with some areas exhibiting strong surface scattering (sky blue area); while ship targets exhibit all three scattering components simultaneously, with relatively high intensity. The composition of the ship's scattering components is also related to its orientation relative to the radar beam. When the ship is perpendicular to the radar beam, strong secondary scattering occurs at the ship's edges. These significant differences provide a reliable basis for distinguishing between ships and chaff based on scattering mechanisms.

[0102] Step 204: Obtain the foreground mask in the SAR image corresponding to the fully polarized data, and apply the foreground mask to the pseudo-color image to obtain the image to be classified.

[0103] Figures 4A to 4C An exemplary fully polarimetric SAR image is shown, in which, Figure 4A For HH channel images, Figure 4B For HV channel images, Figure 4C The image shows the VV channel (the VH channel image is omitted here because it is symmetrical to the HV channel). As can be seen from the image, there is a ship in the plane, and the chaff interference it released has not yet completely separated from the ship.

[0104] In some embodiments, the total power map of the SAR image can be obtained first, where Span = |HH|. 2 +|HV| 2 +|VH| 2 +|VV| 2 ,like Figure 5 As shown; next, a threshold is set based on the target features, and this threshold is used to divide the Span... Figure 2 In binarization, the region is marked as foreground (white) and the sea background is suppressed (black). Then, by removing isolated points from the binarized result, a foreground mask is obtained, as shown below. Figure 6 As shown, white represents ship targets or chaff interference, and black represents the sea background.

[0105] In one example, the CFAR detection method can be used to binarize the span image, where the false alarm rate can be set to, for example, 10. -4 Then, morphological opening operations are used to remove isolated points from the detection results.

[0106] The foreground mask obtained in this way eliminates small isolated noise points and smooths the edges of the detection area.

[0107] Step 205: Classify the pixels in the image to be classified based on the preset scattering feature differences, and label them as ship category pixels and foil category pixels.

[0108] In step 204, a foreground mask is applied to the pseudo-color image, setting the ocean background to black. The only parts to be distinguished are ships and chaff. By using pre-defined differences in scattering characteristics between ships and chaff, such as the proportion of secondary scattering power, volume scattering power, and polarization ratio, each pixel is determined to belong to either the ship or chaff category, thus achieving ship and chaff segmentation in the SAR image. The segmentation result is as follows: Figure 7 As shown, the red area represents ship targets, the green area represents chaff interference, and the black area represents the sea background.

[0109] The method for segmenting ships and chaff in SAR images proposed in this disclosure constructs a coherence matrix for each pixel based on the full polarization data of the SAR radar, and decomposes the coherence matrix into the following scattering mechanisms: surface scattering, secondary scattering, and volume scattering. Then, a pseudo-color image is generated based on the power proportion of each scattering mechanism. Next, a foreground mask of the SAR image is applied to the pseudo-color image to obtain the image to be classified. Finally, the pixels in the image to be classified are classified based on preset scattering feature differences and labeled as ship category pixels and chaff category pixels. This scheme achieves high-precision differentiation between ship and chaff interference through the synergistic effect of polarization scattering mechanism decomposition and mask noise reduction.

[0110] In some embodiments, a Support Vector Machine (SVM) can be used to perform binary classification of images to be classified.

[0111] For sample data in a p-dimensional feature space, SVM achieves class separation by constructing a p-1 dimensional optimal classification hyperplane. When multiple hyperplanes exist that can linearly separate the two classes of samples, the hyperplane that maximizes the margin between the points closest to the hyperplane in both classes is selected as the decision boundary. This optimization objective is equivalent to solving for the hyperplane parameters that maximize the geometric margin under constraints, and the sample points closest to the hyperplane are called support vectors, which directly determine the final position of the classification boundary.

[0112] Suppose we have an l-dimensional training sample dataset {x} i Let i = 1, 2, ..., N. Divide these samples into two classes, m1 and m2. The hyperplane can be obtained using the training samples:

[0113] g(x) = w T x+w0 (10)

[0114] Where w = [w1, w2, ..., w l ] T It is a weighted vector, and w0 is the threshold.

[0115] We need to find a classification hyperplane such that the distances from classes m1 and m2 to the hyperplane are maximized. Since the distances from the sample data to the hyperplane are determined by a weighted vector, i.e.:

[0116] d=|g(x)| / ‖w‖ (11)

[0117] Therefore, to maximize d, we only need to minimize ||w||. Under the following constraints, the Lagrange multiplier method is used to minimize the risk function J(w).

[0118] y(w T x+w0)≥1

[0119] stJ(w)=1 / 2‖w‖ 2 (12)

[0120] For class m1, y = 1, and for class m2, y = -1.

[0121] Due to the small sample size, using SVM for classification can maintain high generalization and stability even in small sample scenarios. For example, in this embodiment, after performing three-component decomposition on 20 ships and 20 chaff, an SVM classifier is trained using the entire pixel set to achieve segmentation of the ships and chaff. The segmentation results are shown in [link to example]. Figure 7 .

[0122] Since the segmentation results obtained above may contain some pixels in the ship region that are identified as chaff, and conversely, the chaff region may contain some pixels that are identified as ships, the segmentation results can be filtered to further optimize the segmentation effect.

[0123] In some embodiments, filtering of the segmentation results can be achieved through the following methods.

[0124] First, obtain the image to be filtered, where each pixel has a label value. Assume the labels for the sea background, ships, and chaff are 0, 1, and 2, respectively, corresponding to the colors black, red, and green.

[0125] Next, for each non-zero pixel (i,j), a fixed-size neighborhood window is selected (usually a 5x5 or 7x7 window). Within this window, the pixel with the most frequent non-zero value is found in its connected neighborhood (surrounding pixels), and this value is used to update the current pixel. Figure 8 For example, if there are more red non-zero pixels in the neighborhood with a value of 0, then the current pixel is updated to red; and so on. Figure 9For example, some pixels in the neighborhood are not connected to the current pixel. Although there are more non-zero green pixels in the neighborhood, there are more red pixels in the connected neighborhood. Therefore, the current pixel is updated to red.

[0126] The filtered result is as follows Figure 10 As shown in the figure, the segmentation results are more uniform and concentrated. In this scheme, analyzing the spatial context information of each pixel's surrounding neighborhood helps correct misclassification, effectively removing isolated misclassified points and smoothing the transition between regions, thus obtaining a more uniform and concentrated segmentation result. The final output clearly distinguishes between the ship and the chaff, and misclassified pixels are basically eliminated, highlighting the effectiveness of the window filtering step in optimizing the segmentation effect.

[0127] Figure 11 This is a block diagram of an exemplary embodiment of a device for segmenting ships and chaff in a SAR image. Figure 11 As shown, the device includes:

[0128] Acquisition unit 1101 is used to acquire the full polarization data of SAR radar;

[0129] The decomposition unit 1102 is used to construct the coherence matrix of each pixel based on the fully polarized data, and decompose the coherence matrix into the following scattering mechanisms: surface scattering, secondary scattering and volume scattering;

[0130] The generation unit 1103 is used to generate a pseudo-color image based on the power ratio of each scattering mechanism;

[0131] Mask unit 1104 is used to acquire the foreground mask in the SAR image corresponding to the fully polarimetric data, and apply the foreground mask to the pseudo-color image to obtain the image to be classified.

[0132] The segmentation unit 1105 is used to classify the pixels in the image to be classified based on the differences in preset scattering features, and label them as ship category pixels and foil category pixels to obtain the segmentation result.

[0133] In some embodiments, the decomposition unit is used for:

[0134] The coherence matrix is ​​equivalent to a weighted combination of planar scattering power, secondary scattering power, and volume scattering power.

[0135] The total power of a SAR image is equal to the sum of the plane scattering power, the secondary scattering power, and the volume scattering power, along with the weight values ​​corresponding to the three types of scattering power, and each scattering power is solved.

[0136] In some embodiments, the mask unit is used for:

[0137] Obtain the total power map of the SAR image;

[0138] A threshold is set based on the target features, and the total power map is binarized using the threshold.

[0139] The isolated points in the binarization result are removed to obtain the foreground mask.

[0140] In some embodiments, when the mask unit is used to set a threshold based on target features and to binarize the total power map using the threshold, it is specifically used to: perform constant false alarm rate detection on the total power map for binarization;

[0141] The mask unit is used to obtain the foreground mask by removing isolated points from the binarization result, specifically as follows:

[0142] The foreground mask is obtained by removing isolated points from the binarization result through morphological opening operations.

[0143] In some embodiments, classifying the pixels in the image to be classified based on preset scattering feature differences and labeling them as ship category pixels and chaff category pixels includes:

[0144] The pixels in the image to be classified are classified using a support vector machine, wherein the support vector machine is trained using the power proportion values ​​of the scattering mechanisms of multiple ships and multiple foil strips.

[0145] In some embodiments, the device further includes a filtering unit for:

[0146] The image to be filtered is obtained based on the label value of each pixel in the segmentation result, wherein the label value of the background pixel is zero and the label value of the foreground pixel is non-zero.

[0147] For each non-zero pixel, select a neighborhood window of a fixed size;

[0148] Within the neighborhood window, the current pixel is updated using the most non-zero label value in the connected neighborhood of the current non-zero pixel.

[0149] Figure 12 This is a schematic structural diagram of a device provided in an exemplary embodiment. Please refer to... Figure 12At the hardware level, the device includes a processor 1202, an internal bus 1204, a network interface 1206, memory 1208, and non-volatile memory 1210, and may also include other hardware required for business operations. One or more embodiments of this specification can be implemented in software, such as the processor 1202 reading the corresponding computer program from the non-volatile memory 1210 into memory 1208 and then running it. Of course, in addition to software implementation, one or more embodiments of this specification do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0150] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer, which can take the form of a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email sending and receiving device, game console, tablet computer, wearable device, or any combination of these devices.

[0151] In a typical configuration, a computer includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0152] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0153] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0154] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0155] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0156] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of one or more embodiments of this specification. The singular forms “a,” “described,” and “the” used in one or more embodiments of this specification and in the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0157] It should be understood that although the terms first, second, third, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of one or more embodiments of this specification, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "in response to a determination," or "when," or "in the event of a determination."

[0158] The above description is merely a preferred embodiment of one or more embodiments of this specification and is not intended to limit the scope of one or more embodiments of this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments of this specification should be included within the protection scope of one or more embodiments of this specification.

Claims

1. A method for segmenting ships and chaff in SAR images, characterized in that, include: Acquire full polarization data from SAR radar; The coherence matrix of each pixel is constructed based on the fully polarized data, and the coherence matrix is ​​decomposed into the following scattering mechanisms: surface scattering, secondary scattering, and volume scattering; Pseudo-color images are generated based on the power proportion of each scattering mechanism; Obtain the foreground mask in the SAR image corresponding to the fully polarimetric data, and apply the foreground mask to the pseudo-color image to obtain the image to be classified; The pixels in the image to be classified are classified based on the differences in preset scattering features and labeled as ship category pixels and chaff category pixels to obtain the segmentation result; The coherence matrix is ​​decomposed into the following scattering mechanisms: surface scattering, secondary scattering, and volume scattering, including: The coherence matrix is ​​equivalent to a weighted combination of planar scattering power, secondary scattering power, and volume scattering power. The total power of a SAR image is equal to the sum of plane scattering power, secondary scattering power, and volume scattering power, along with the weight values ​​corresponding to the three types of scattering power. Each scattering power is then solved. The weight values ​​of plane scattering power, secondary scattering power, and volume scattering power are respectively the coherence matrix templates for surface scattering, secondary scattering, and volume scattering. The coherence matrix templates are predefined based on the physical scattering model.

2. The method according to claim 1, characterized in that, The step of obtaining the foreground mask in the SAR image corresponding to the fully polarimetric data includes: Obtain the total power map of the SAR image; A threshold is set based on the target features, and the total power map is binarized using the threshold. The isolated points in the binarization result are removed to obtain the foreground mask.

3. The method according to claim 2, characterized in that, The step of setting a threshold based on target features and using the threshold to binarize the total power map includes: performing constant false alarm rate detection on the total power map to perform binarization; The process of removing isolated points from the binarization result to obtain the foreground mask includes: The foreground mask is obtained by removing isolated points from the binarization result through morphological opening operations.

4. The method according to claim 1, characterized in that, The step of classifying pixels in the image to be classified based on preset scattering feature differences and labeling them as ship category pixels and chaff category pixels includes: The pixels in the image to be classified are classified using a support vector machine, wherein the support vector machine is trained using the power proportion values ​​of the scattering mechanisms of multiple ships and multiple foil strips.

5. The method according to claim 1, characterized in that, The method further includes: The image to be filtered is obtained based on the label value of each pixel in the segmentation result, wherein the label value of the background pixel is zero and the label value of the foreground pixel is non-zero. For each non-zero pixel, select a neighborhood window of a fixed size; Within the neighborhood window, the current pixel is updated using the most non-zero label value in the connected neighborhood of the current non-zero pixel.

6. A device for segmenting ships and chaff in SAR images, characterized in that, include: Acquisition unit, used to acquire full polarization data of SAR radar; The decomposition unit is used to construct the coherence matrix of each pixel based on the fully polarized data, and decompose the coherence matrix into the following scattering mechanisms: surface scattering, secondary scattering and volume scattering; The generation unit is used to generate pseudo-color images based on the power proportion of each scattering mechanism; A masking unit is used to acquire the foreground mask in the SAR image corresponding to the fully polarimetric data, and apply the foreground mask to the pseudo-color image to obtain the image to be classified. The segmentation unit is used to classify the pixels in the image to be classified based on the differences in preset scattering features, and label them as ship category pixels and chaff category pixels to obtain the segmentation result; The coherence matrix is ​​decomposed into the following scattering mechanisms: surface scattering, secondary scattering, and volume scattering, including: The coherence matrix is ​​equivalent to a weighted combination of planar scattering power, secondary scattering power, and volume scattering power. The total power of a SAR image is equal to the sum of plane scattering power, secondary scattering power, and volume scattering power, along with the weight values ​​corresponding to the three types of scattering power. Each scattering power is then solved. The weight values ​​of plane scattering power, secondary scattering power, and volume scattering power are respectively the coherence matrix templates for surface scattering, secondary scattering, and volume scattering. The coherence matrix templates are predefined based on the physical scattering model.

7. An electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor implements the method as described in any one of claims 1-5 by executing the executable instructions.

8. A computer-readable storage medium having stored thereon computer instructions that, when executed by a processor, implement the steps of the method as claimed in any one of claims 1-5.

9. A computer program product comprising a computer program / instructions that, when executed by a processor, implement the steps of the method as claimed in any one of claims 1-5.

Images