A ship detection method for full polarimetric SAR image

Abstract

The application discloses a full polarization SAR image ship detection method, comprising the following steps: S1, pre-processing a full polarization SAR image to obtain different types of scattering power and polarization wave anisotropy parameters of each pixel; S2, using the decomposed different scattering power and combining the polarization wave anisotropy parameters to construct a full polarization SAR image detector; S3, using a generalized gamma distribution to describe the statistical characteristics of the full polarization SAR image detector to realize constant false alarm rate adaptive detection of the full polarization SAR image detector. The scattering characteristics and anisotropy of the scattering wave of the ship target are used to fully represent the ship target, so that the ship is strengthened and the background is weakened, and the ship can be detected from the complex marine background.

Description

Technical Field

[0001] This invention relates to the field of synthetic aperture radar target detection and recognition technology, and more specifically, to a method for ship detection in fully polarimetric SAR images that combines target scattering characteristics and polarimetric anisotropy. Background Technology

[0002] Synthetic aperture radar (SAR) occupies an important position in Earth observation technology due to its unique advantage of being unaffected by weather conditions and capable of imaging day and night in all weather conditions. SAR imagery for ship detection and identification has many potential applications in commerce, fisheries, shipping services, and military sectors, especially in maritime traffic control, illegal fishing, illegal vessel intrusion, and marine environmental monitoring. The main difficulties in SAR image-based ship detection include: (1) the complex imaging mechanism and sea conditions of SAR images, making it difficult to characterize the target and ocean background characteristics; (2) the difficulty in describing the statistical characteristics of ocean clutter using traditional statistical models due to environmental factors such as wind speed and waves; and (3) the difficulty in effectively monitoring small ships against complex ocean backgrounds. SAR images contain a large amount of clutter background, and ships are only a small part of the ocean relative to it. Furthermore, the influence of natural environmental factors such as wind speed and waves makes ship detection extremely difficult.

[0003] In recent years, with the continuous development of satellite technology, from the initial single-polarization mode to the current multi-polarization mode, SAR images contain increasingly rich information, especially in fully polarimetric SAR data, which contains abundant polarimetric information and has played a significant advantage in ship detection. However, this rich polarimetric information also means prominent clutter backgrounds, mainly manifested in prominent noise. In this context, traditional detectors such as SPAN and PMS are no longer suitable for ship detection in fully polarimetric SAR images, necessitating a ship detector specifically designed for these images. The DBSP detector, as an excellent SAR image ship detector in recent years, considers the information of the eight neighborhoods surrounding each pixel, constructs a polarimetric covariance difference matrix for each pixel, and then uses this matrix for four-component decomposition, decomposing the total power into the sum of specular scattering, double-hop scattering, volume scattering, and spiral scattering power. For the DBSP detector, the sum of double-hop scattering and volume scattering is multiplied by spiral scattering, and specular scattering is discarded, thus forming the DBSP detector. The DBSP detector has shown excellent detection performance on fully polarimetric SAR data, but some shortcomings still remain. For example, the DBSP detector only considers the scattering characteristics of the ship; in other words, DBSP is a scattering-dominant method. On the one hand, when a ship has a simple structure and mainly exhibits specular scattering, the DBSP detector is prone to missing detections of such ships. On the other hand, the construction of the polarization covariance difference matrix depends on the difference between the central pixel and its eight neighboring pixels. That is, when the scattering difference between the ship pixel and its surrounding pixels is not significant, the value of the central pixel may be very small, ultimately leading to a very small detector value and thus losing the ability to detect the ship.

[0004] Therefore, it is necessary to develop a fully polarimetric SAR image ship detection method that combines target scattering characteristics and polarimetric wave anisotropy to overcome the above-mentioned technical problems. Summary of the Invention

[0005] The purpose of this invention is to provide a ship detection method for fully polarimetric SAR images to overcome the shortcomings of existing technologies.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for ship detection using fully polarimetric SAR images includes the following steps:

[0008] S1. Preprocess the fully polarimetric SAR image to obtain the scattering power and polarization anisotropy parameters of different types of each pixel;

[0009] S2. Construct a fully polarimetric SAR image detector by utilizing the different scattering powers obtained from the decomposition and combining them with the polarization wave anisotropy parameters.

[0010] S3. The statistical characteristics of the fully polarimetric SAR image detector are described by the generalized gamma distribution, so as to realize the constant false alarm rate adaptive detection of the fully polarimetric SAR image detector.

[0011] Further, step S1 includes:

[0012] S10. Extract the polarization information of the fully polarimetric SAR image, calculate the polarization covariance matrix, and then use the covariance matrix to perform four-component decomposition of the scattering power of each pixel.

[0013] S11. Using a two-dimensional complex scattering matrix, construct the Mueller matrix, then solve for the maxima and minima of the polarizability, and use the extreme values ​​of the polarizability to obtain the anisotropy of the polarized wave.

[0014] Further, step S10 specifically includes:

[0015] Utilizing polarization information Find the polarization covariance matrix C, where C = k·k *T Next, the scattering power is decomposed into four components using the covariance matrix, and the decomposition formula is as follows:

[0016]

[0017] The parameters a, b, c, d, and e in the formula are derived from... If determined, then P s =f s (1+|β| 2 ), P d =f d (1+|α| 2 ), P v =f v P h =f h And P s P d P v and P h Satisfy P t =P s +P d +P v +P h =SPAN;

[0018] Further, step S11 specifically includes:

[0019] First, using a two-dimensional complex scattering matrix For each cell, construct the corresponding Mueller matrix. The expression for the Mueller matrix is ​​as follows:

[0020]

[0021] M = RWR -1

[0022] in,

[0023] Then, the extreme values ​​of polarizability, i.e., p, are obtained using the Mueller matrix. min and p max The extreme values ​​of polarizability are obtained using S(p) = -ln(s(p)), where

[0024] Finally, using ΔS n =S n (p min )-S n (p max ), to obtain the value of polarization wave anisotropy.

[0025] Furthermore, step S3 specifically includes:

[0026] S31. Using the generalized gamma distribution to describe the statistical characteristics of clutter background;

[0027] S32. The threshold for detection is estimated using the generalized gamma distribution to achieve adaptive detection of fully polarimetric SAR images.

[0028] Further, step S32 specifically includes:

[0029] The generalized gamma distribution GΓD is used to characterize the statistical behavior of the fully polarimetric SAR image detector in marine scenes, so as to achieve adaptive CFAR detection in different scenes. The probability density function formula of the generalized gamma distribution GΓD is as follows:

[0030]

[0031] Where σ, v represents the scale parameter, power, shape, and Γ(·) represents the Gamma function.

[0032] Furthermore, the threshold calculation formula for the detection quantity in step S32 is as follows:

[0033]

[0034] Among them, Γ -1 (·,·) denotes the inverse incomplete function, P fa This is a given value for the false alarm probability.

[0035] Compared with the prior art, the advantages of the present invention are: the present invention utilizes the scattering characteristics of ship targets and the anisotropy of scattered waves to fully characterize ship targets, so as to enhance the ship and weaken the background, so as to make it possible to detect the ship from the complex ocean background. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating the ship detection method for fully polarimetric SAR images of the present invention.

[0038] Figure 2 This is the fitting diagram of the generalized gamma distribution clutter in this invention.

[0039] Figure 3 In this invention, a is the Pauli decomposition map of a large scene, and b is the joint-SA feature map.

[0040] Figure 4 In the diagram, a is the Pauli decomposition diagram in this invention, and b is the three-dimensional visualization of the region.

[0041] Figure 5 In this invention, 'a' represents the feature map corresponding to the detection quantity, and 'b' represents the detection quantity.

[0042] Figure 6 In this paper, a represents the detection result of the DBSP detector and b represents the detection result of the joint-SA detector. Detailed Implementation

[0043] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby providing a clearer and more explicit definition of the scope of protection of the present invention.

[0044] See Figure 1 As shown, in order to accurately detect ship targets in complex marine backgrounds and overcome the shortcomings of existing detectors that cannot effectively characterize the characteristics of ship targets, this embodiment mainly considers the scattering characteristics and anisotropy of the scattered waves of ship targets. First, the scattering power of ship targets is decomposed into four components (specular scattering, double-hop scattering, volume scattering, and spiral scattering). Then, the anisotropy value of the scattered waves is solved using a two-dimensional complex matrix. Finally, the retained scattering power is multiplied by the anisotropy value using the multiplicative rule to obtain the final detection quantity that can describe the characteristics of the ship.

[0045] This embodiment discloses a method for ship detection in fully polarimetric SAR images that combines target scattering characteristics and polarimetric anisotropy, specifically including the following steps:

[0046] Step S1: Preprocess the fully polarimetric SAR image to obtain the scattering power and polarization anisotropy parameters of different types of each pixel. Specifically, this includes: First, extracting the polarization information of the fully polarimetric SAR image and calculating the polarization covariance matrix. Then, using the covariance matrix, performing a four-component decomposition of the scattering power of each pixel. Next, using the two-dimensional complex scattering matrix, constructing the Mueller matrix, and then solving for the maximum and minimum values ​​of the polarimetric degree. The polarization anisotropy is obtained using the extreme values ​​of the polarimetric degree.

[0047] In this embodiment, the C-band fully polarimetric SAR image data acquired by the Chinese Gaofen-3 satellite is first preprocessed, mainly through filtering. Fully polarimetric SAR measures complete backscattering information of the target, comprehensively characterizing the target's scattering power. Typically, a two-dimensional complex scattering matrix [S] is used to characterize the information of each pixel unit in the image, defined as:

[0048]

[0049] Or it can be equivalently represented as a scattering vector:

[0050]

[0051] In the formula, Trace(·) is the sum of the diagonal elements of the matrix, which is the complete set of 2x2 basis matrices under the Hermitian inner product. Finally, the scattering mechanism can be defined as a normalized vector. ω =k / | k |

[0052] Typically, targets observed by SAR are not ideal scattering mechanisms, but rather combinations of different objects, generally referred to as partial targets. To characterize a local target, a single scattering matrix is ​​insufficient; therefore, this embodiment introduces second-order statistics of the scattering matrix. In this case, the target's covariance matrix can be estimated, assuming that this scattering vector matrix satisfies the reciprocity theorem, i.e. The covariance matrix can then be expressed as:

[0053]

[0054] Where <·> denotes spatial average, and * denotes complex conjugation. The power of the covariance matrix is ​​defined as:

[0055] SPAN=C 11 +C 12 +C 13

[0056] Next, the covariance matrix is ​​used to decompose each pixel into four components, and the anisotropy of the scattered wave is solved using the two-dimensional complex scattering matrix.

[0057] First, the covariance matrix C is decomposed into four components, and the decomposition expression is as follows:

[0058]

[0059] The parameters a, b, c, d, and e in the formula are derived from... To determine, then P s =f s (1+|β| 2 ), P d =f d (1+|α| 2 ), P v =f v P h =f h And P s P d P v and P h Satisfy P t =P s +P d +P v +P h =SPAN.

[0060] Second, solve for the anisotropy of the scattered waves:

[0061] First, using the two-dimensional complex scattering matrix, the Muller matrix is ​​constructed, and its expression is shown in the following formula:

[0062]

[0063] M = RWR -1

[0064] in,

[0065] Then, the extreme values ​​of polarizability, i.e., p, are obtained using the Mueller matrix. min and p max .

[0066] Furthermore, an entropy value S for polarized waves is introduced, which uses polarizability to measure the entropy of the polarized wave. Its expression is as follows:

[0067] S(p)=-ln(s(p))

[0068] in, The formula shows that the entropy S of a polarized wave depends only on the polarizability p and is independent of the polarization state, satisfying S(P=1)≤S≤S(P=0). When the scattered wave is fully polarized, the entropy S is 0, while when the scattered wave is fully unpolarized, it is 1, i.e., P=0. Entropy ΔS n The dynamic range of the scattered wave is directly related to the dynamic range of the polarizability Δp, thus providing a measure of the complexity of the scattering mechanism. This leads to the expression for the polarization anisotropy of the scattered wave:

[0069] ΔS n =S n (p min )-S n (p max )

[0070] Therefore, the above equation can be used to characterize the nonstationarity of the target: ΔS n The higher the value, the greater the change in signal polarization with transmission and reception, and this metric has been proven to be an effective tool for improving ship-to-sea contrast.

[0071] Step S2: Using the decomposed different scattering powers and combining them with the polarization wave anisotropy parameters, a fully polarimetric SAR image detector is constructed, specifically as follows:

[0072] In step S1, the four types of scattering power (spectral scattering P) have already been solved. s Double-hop scattering P d Volume scattering P v Helical scattering P h Given the values ​​of the anisotropy of the scattered wave and the scattering wave, a new detector can be constructed, expressed as follows:

[0073] joint-SA=(P d +P v )·P h ·ΔS

[0074] Based on this detection data, the characteristics of ship targets and sea clutter are visualized and described, such as... Figure 3 As shown in the image, the bright areas represent the corresponding ships, while the darker areas represent the sea clutter background. It can be seen that using the aforementioned detection methods not only effectively highlights ship targets but also provides significant suppression of sea clutter.

[0075] To further demonstrate the advantages of the aforementioned detector, this embodiment selects a small region containing only one weakly scattering target, such as... Figure 4 As shown, the area is visualized in three dimensions. Figure 4 (a) is a Pauli exploded view, from which it can be seen that the bright area in the middle is the ship target; Figure 4(b) is a three-dimensional visualization of the area. The raised part in the middle represents the weakly scattering target. It can be clearly seen that the values ​​of the clutter background are all between 0 and 0.1, while the peak value of the weakly scattering target is around 0.6. Therefore, it can be proven that the detection quantity in this invention is very effective in highlighting the weakly scattering target and suppressing the background.

[0076] Step S3: Use the generalized gamma distribution to describe the statistical characteristics of the fully polarimetric SAR image detector, and realize the constant false alarm rate adaptive detection of the fully polarimetric SAR image detector.

[0077] After defining the fully polarimetric SAR image detector, a distribution model is needed to characterize the statistical behavior of the new detector in marine scenes to achieve adaptive CFAR detection in different scenarios. In this embodiment, the statistical distribution of GTD is used to characterize the statistics of joint-SA. GTD is used because it has been widely applied in various fields and has proven to be an effective method for describing the statistical behavior of sea clutter; its probability density function (PDF) is:

[0078]

[0079] Where, σ, v represents the scale parameter, power, shape, and Γ(·) represents the Gamma function. Although the gamma distribution is an empirical model, it has been shown to be relatively general. Therefore, the detection threshold can be expressed as:

[0080]

[0081] Among them, Γ -1 (·,·) denotes the inverse incomplete function, P fa This is a given value for the false alarm probability.

[0082] from Figure 2 As can be seen, under the detection parameters of this invention, the generalized gamma distribution can effectively describe the statistical behavior characteristics of sea clutter, and therefore can be used to adaptively estimate the detection parameters in this invention. After estimating the threshold using the generalized gamma distribution, adaptive CFAR detection is performed using this threshold, such as... Figure 5 As shown, Figure 5 (a) is the feature map corresponding to the detection quantity in this invention. Figure 5 (b) The detection results using the detection parameters of this invention and the corresponding CFAR adaptive detection technology. As can be seen from the figure, for the detection parameters of this invention, the CFAR adaptive detection technology can effectively detect ships in the selected area without missed detections or false alarms.

[0083] To further illustrate the superiority of the detection quantity in this invention, the detection quantity in this invention is compared with other detection quantities, such as... Figure 6 As shown. Figure 6 (a) shows the detection results of the DBSP detector. As can be seen from the figure, although the DBSP successfully detected the ship, it also resulted in missed detections and false alarms. In contrast, the detection results of this invention successfully detected the ship without any false alarms or missed detections. Therefore, it can be proven that the detection method of this invention is superior to other detection methods.

[0084] This invention fully utilizes the scattering characteristics of ship targets. More specifically, the ocean background exhibits more specular scattering, while ship targets exhibit more double-hop scattering, volume scattering, and spiral scattering. Theoretically, it can be explained that spiral scattering exists only in ship targets. Therefore, this invention can utilize spiral scattering as an enhancement factor to increase the scattering power of ship targets, thereby highlighting the ship targets and weakening the ocean background.

[0085] This invention fully considers the polarization characteristics of scattered waves. Compared to ship targets, the ocean background has a more stable surface, thus exhibiting more specular scattering from a scattering perspective, and thus showing more isotropic polarization. For ship targets, however, due to their relatively complex structure and rougher surface, the scattered waves exhibit more anisotropic polarization.

[0086] Therefore, by utilizing the scattering characteristics of ship targets and the anisotropy of scattered waves, this invention can fully characterize ship targets, thereby enhancing the ship and weakening the background, and enabling the detection of ships from complex ocean backgrounds.

[0087] Although embodiments of the present invention have been described in conjunction with the accompanying drawings, the patent owner may make various modifications or alterations within the scope of the appended claims, as long as they do not exceed the protection scope described in the claims of the present invention, they shall be within the protection scope of the present invention.

Claims

1. A method for ship detection using fully polarimetric SAR images, characterized in that, Includes the following steps: S1. Preprocess the fully polarimetric SAR image to obtain the scattering power and polarization anisotropy parameters of different types of each pixel; S2. Construct a fully polarimetric SAR image detector by utilizing the different scattering powers obtained from the decomposition and combining them with the polarization wave anisotropy parameters. S3. The statistical characteristics of the fully polarimetric SAR image detector are described by the generalized gamma distribution, so as to realize the constant false alarm rate adaptive detection of the fully polarimetric SAR image detector. Step S1 includes: S10. Extract the polarization information of the fully polarimetric SAR image, calculate the polarization covariance matrix, and then use the covariance matrix to perform four-component decomposition of the scattering power of each pixel. S11. Using the two-dimensional complex scattering matrix, construct the Mueller matrix, then solve for the maximum and minimum values ​​of the polarizability, and use the extreme values ​​of the polarizability to obtain the anisotropy of the polarized wave. Step S10 specifically involves: Utilizing polarization information Find the polarization covariance matrix C, where C = k·k* T Next, the scattering power is decomposed into four components using the covariance matrix, and the decomposition formula is as follows: The parameters a, b, c, d, e in the formula are derived from... If determined, then P s =f s (1+|β| 2 ), P d =f d (1+|α| 2 ), P v =f v P h =f h And P s P d P v and P h Satisfy P t =P s +P d +P v +P h =SPAN.

2. The ship detection method for fully polarimetric SAR images according to claim 1, characterized in that, Step S11 specifically involves: First, using a two-dimensional complex scattering matrix For each cell, construct the corresponding Mueller matrix. The expression for the Mueller matrix is ​​as follows: M=RWR -1 in, Then, the extreme values ​​of polarizability, i.e., p, are obtained using the Mueller matrix. min and P max The extreme values ​​of polarizability are obtained using S(p) = -ln(s(p)), where Finally, using ΔS n =S n (p min )-S n (p max ), to obtain the value of polarization wave anisotropy.

3. The ship detection method for fully polarimetric SAR images according to claim 1, characterized in that, Step S3 specifically includes: S31. Using the generalized gamma distribution to describe the statistical characteristics of clutter background; S32. Use the generalized gamma distribution to estimate the threshold of the detection quantity to achieve adaptive detection of fully polarimetric SAR images.

4. The ship detection method for fully polarimetric SAR images according to claim 3, characterized in that, Step S32 specifically involves: The generalized gamma distribution GΓD is used to characterize the statistical behavior of the fully polarimetric SAR image detector in marine scenes, so as to achieve adaptive CFAR detection in different scenes. The probability density function formula of the generalized gamma distribution GΓD is as follows: Where σ, v represents the scale parameter, power, shape, and Γ(·) represents the Gamma function.

5. The ship detection method for fully polarimetric SAR images according to claim 3, characterized in that, The formula for calculating the threshold of the detection quantity in step S32 is as follows: Among them, Γ -1 (·,·) denotes the inverse incomplete function, P fa This is a given value for the false alarm probability.

Images