For circular polarization transmitting linear polarization receiving (CTLR) compact polarimetric (CP) synthetic aperture radar (SAR) data, the volume scattering power proportion derived through CP decomposition, such as the$m\text {-}\alpha $decomposition algorithm, is typically overestimated. To address this issue, this study initially presents a theoretical analysis of the reason behind the overestimation of the volume scattering power proportion. The theoretical analysis reveals that the assumption “the volume scattering power proportion is$1\text {-}m$” employed by the$m-\alpha $decomposition indeed leads to an overestimation, where m represents the degree of polarization. Subsequently, a novel and smaller form of the volume scattering power proportion has been discovered, namely ($1\text {-}m)^{2}$. Based on this finding, a modified$m\text {-}\alpha $decomposition algorithm has been proposed. Experiments were conducted using two fully polarimetric SAR images and a CP SAR image acquired by E-SAR, GF-3, and RISAT. The experimental results demonstrate that: 1) ($1\text {-}m)^{2}$is statistically closer to the actual volume scattering power proportion than$1\text {-}m$; 2) the decomposition performance of the modified$m\text {-}\alpha $decomposition algorithm surpasses that of the other six compact decomposition algorithms; and 3) employing the new volume scattering power proportion ($1\text {-}m)^{2}$effectively mitigates the issue of volume scattering power overestimation in CP decomposition.

Based on the input schemes in Polarimetric Synthetic Aperture Radar (PolSAR) classification through deep learning. In this paper, we propose eight polarimetric parameter input schemes through reflection symmetric decomposition (RSD) and adopt classic convolutional neural networks for the classification of GF-3 PolSAR data.The results demonstrate that the commonly used 6-parameter input scheme, which many researchers adopt, lacks the comprehensive utilization of polarization information and warrants attention. The 7 alternative methods designed in addition to this scheme are all superior to it. If the computational resources are limited, it is suggested to directly use the scheme 7, which includes all the information of the T matrix. If the device configuration allows, it is recommended to prioritize the use of the 21-parameter polarization data input scheme 8, which includes all the parameters of the T matrix and the RSD.

Ship target search tasks aim to match specific ships across two or more satellite images. Like pedestrian and vehicle re-identification tasks in computer vision, accurate ship re-identification encounters challenges, including subtle differences between ships of the same type and substantial intra-instance variations due to satellite angle of view and spectral differences. To tackle these challenges, this paper introduces a deep learning-based two-branch framework for ship target search, integrating ship detection and re-identification tasks. One branch extracts the target ship features while the other captures the search region features. These features are then fused through a dedicated layer, and the final output is derived from the keypoint detection header. A new dataset was curated using Sentinel-2 and Gaofen-1 satellite data. Experimental results validate the robustness of the proposed method, achieving an accuracy of 94.37% on the new dataset. Our method’s scalability has been validated through experiments using CBERS-04 and Gaofen-6 satellite data.

Ship detection in synthetic aperture radar (SAR) images using a convolutional neural network (CNN) has become a hotspot. However, dense multi-size ship targets in inshore scenes, proximity of ship targets to man-made targets, and datasets containing only one class limit the performance of ship detection methods. In pursuit of attaining heightened performance, most existing methods endeavor to augment the number of parameters and enhance network complexity. To address these problems, this letter proposes an anchor-free ghost feature extraction and cross-scale interaction network (GFECSI-Net), which improves detection performance through an efficient implementation, thereby avoiding the increase in the number of parameters or network complexity. First, to enhance the capability of feature extraction for ship targets, a multi-scale adaptive feature pyramid network (MSAFPN) is proposed to realize intensive information interaction and cross-scale feature fusion between different feature maps. Meanwhile, a selective efficient channel attention module (SECAM) is designed to enable the network to prioritize channels that better characterize ship targets. Besides, a GPU-efficient backbone for generating ghost feature maps and a task alignment detection head are integrated into GFECSI-Net. Comparison results with nine state-of-the-art CNN methods on a high-resolution SAR image dataset (HRSID) and a large-scale multi-class SAR target dataset (MSAR-1.0) indicate that GFECSI-Net achieves superior detection performance in F1-Score and mAP with a small number of parameters.

Model-based Incoherent Polarimetric Decomposition (MIPD) is a frequently used technique to analyze the scattering characteristics of multilook Polarimetric Synthetic Aperture Radar (PolSAR) data. A reflection symmetry decomposition (RSD) algorithm was proposed to overcome two limitations of Freeman–Durden decomposition. In this letter, a modified RSD (MRSD) algorithm with an improved approach to determining the surface scattering power and the double-bounce scattering power of RSD is proposed. MRSD causes no loss of polarimetric information, and its outputs consist of nine independent real-valued parameters. Based on these two properties, a new polarimetric product is created for Chinese GF-3 satellite. The product is very convenient for PolSAR data application users because it can be directly displayed by geographic information system (GIS) software, and its ground pixel sizes are very close to be square. The case study with six real GF-3 images demonstrates the effectiveness of both MRSD and the new product.

Model-based incoherent polarimetric decomposition is a frequently used technique to analyze multilook data of polarimetric synthetic aperture radars (POLSARs). The purpose of this study is to analyze and compare different model-based incoherent polarimetric decomposition algorithms from the polarimetric information change aspect. For the input of a model-based incoherent polarimetric decomposition algorithm, polarimetric entropy was used to represent the polarimetric information of a coherency matrix. For the output of a model-based incoherent polarimetric decomposition algorithm, there are usually several decomposed components. To quantitatively represent their total polarimetric information, a new concept, generalized polarimetric entropy, was proposed which generalized the concept of polarimetric entropy based on the information entropy additivity of information theory. Generalized polarimetric entropy consists of two parts named as polarimetric power entropy and polarimetric residual entropy, respectively. Polarimetric power entropy describes the distribution status of the Span values of all decomposed components. Polarimetric residual entropy represents the residual randomness of all decomposed components. With the three new concepts, eight model-based incoherent polarimetric decomposition algorithms were compared and analyzed. Two real POLSAR images, respectively, derived by the E-SAR airborne system of Germany and the GF-3 satellite of China were used for the experiments. Experimental results had illustrated several useful conclusions.

Polarimetric symmetry, including reflection symmetry, rotation symmetry, and azimuthal symmetry, are useful concepts for describing the scattering characteristics of polarimetric synthetic aperture radar (SAR) data. In this study, for the coherency matrices of multilook polarimetric SAR data, three new scattering models were first proposed based on the polarimetric symmetry. Then, a four-component incoherent decomposition algorithm was introduced, which was based on the three aforementioned new models, along with a classic volume scattering model. The proposed algorithm was considered to be a complete incoherent decomposition algorithm. No polarimetric information was lost during the decomposition. Also, there were no negative powers in its output. Then, by utilizing three polarimetric SAR images which had been derived by ESAR, RADARSAT-2, and GF-3, respectively, the experimental results had revealed that the proposed decomposition algorithm was effective in the analysis of the scattering mechanisms of terrain targets. It was observed that more than 90% of the coherency matrices of the real polarimetric SAR data had been completely decomposed by the proposed algorithm into four components. These components were found to be exactly consistent with the scattering models for surface scattering, double-bounce scattering, volume scattering, and helix scattering. The proposed decomposition algorithm had provided insight into the limitations of incoherent decomposition.

Freeman-Durden decomposition is a frequently used technique to analyze the scattering characteristics of multilook Polarimetric Synthetic Aperture Radar (POLSAR) data. When it is applied to real POLSAR data, two problems emerge, which are the volume scattering overestimation and negative powers. Many researchers think these two problems are caused by the insufficient decomposition algorithm, and several improvements are proposed. However, the improved decomposition algorithms become more and more complicated, and some new problems such as the decomposed component is not model-based also emerge. In this article, we try to solve the two problems through another way. We think they are caused by the dogmatic input rather than the insufficient decomposition algorithm. Freeman-Durden decomposition explicitly assumes reflection symmetry. Its input is a direct truncation of the measured coherency matrix. The truncation can be regarded as a Reflection Symmetry Approximation (RSA) of the measured coherency matrix. We firstly show some reasons why we think the truncation is not a good RSA. Then a new RSA is proposed based on the sum of three reflection symmetry components derived from the measured coherency matrix. Experimental results with several real POLSAR images show that, if the new RSA is used as the input of Freeman-Durden decomposition, the abovementioned two problems no longer exist.

The HY-2B satellite is the second dynamic environment satellite in China. It was successfully launched on October 24th 2018 with a sun-synchronous orbit at an altitude of~970km. Repeat cycles of 14 days are planned for the first two years with oceanographic purpose and 168 days geodetic cycles will follow for the third year of the mission. The satellite is equipped with a Ku/C bands altimeter and the orbit is determined thanks to SLR, GPS and DORIS systems.

Freeman-Durden decomposition is a frequently used technique to analyze the scattering characteristics of multilook Polarimetric Synthetic Aperture Radar (POLSAR) data. When it is applied to the real POLSAR data, two problems emerge, namely, the volume scattering overestimation and negative powers. Many researchers think these two problems are caused by the insufficient decomposition algorithm, and several improvements are proposed. However, the improved decomposition algorithms become more and more complicated, and some new problems such as the decomposed component is not model based also emerge. In this paper, we try to solve the two problems through another way. We think they are caused not by the insufficient decomposition algorithm but by the dogmatic input. Freeman-Durden decomposition explicitly assumes reflection symmetry. Its input is a direct truncation of the measured coherency matrix. The truncation can be regarded as a reflection symmetry approximation (RSA) of the measured coherency matrix. We first show some reasons why we do not think the truncation is a good RSA. Then, a new RSA is proposed based on the sum of three reflection symmetry components derived from the measured coherency matrix. Experimental results with several real POLSAR images show that, if the new RSA is used as the input of Freeman-Durden decomposition, the above-mentioned two problems no longer exist.

Nowadays, several unitary transformations have been used in model-based incoherent decomposition of polarimetric synthetic aperture radar data. However, the physical meanings of these unitary transformations are still vague (except the orientation angle compensation). In this paper, we propose calling a unitary transformation as the Helix Angle Compensation (HAC), and detailed analyses on its physical meanings are presented. We show that HAC denotes the structure information of a target that it contains different double-bounce scattering components with both different orientation angles and different distances form radar along the line of radar sight. HAC is used to modify a three-component incoherent decomposition algorithm, and makes its third component exactly consistent with the surface scattering model or the double-bounce scattering model. Experiments with real polarimetric SAR date are presented to demonstrate the effectiveness of the modified decomposition and illustrate several properties of HAC.

How to extract oil spill information accurately with full polarization SAR data is the current problem, Polarization decomposition is one of the effective ways to improve the accuracy of oil spill information detection. Based on the polarization decomposition parameter entropy H, scattering Angle and anti-entropy A, this paper analyzes polarization decomposition parameters and carries out extraction of oil spill information. In the light of the similar characteristics of oil spill in entropy H and scattering angle parameters, platform and ship is analyzed, which bring the false of oil spill, and various indexes are constructed. From scattering mechanism and test results, established the index of the oil spill information extraction, combined with the oil spill features, this paper proposes a simple and effective information automatic extraction method of oil spill, The accuracy is about 71.4%, which can meet the needs of oil spill operation monitoring.

Based on multi-temporal observations of static targets on sea surface, a positioning shift that relates to geodetic terrain heights (GTH) is found. The positioning shift mainly exists in the range direction of a SAR image and has an almost linear relationship with the difference between the real GTH of a target and the GTH of a SAR product. An approach of compensating the positioning shift is proposed based on rigorous projection models of RADARSAT-2. Experimental results illustrate that the positioning error of a static target which is derived using SCW products with different GTHs is 446.63m. After applying the proposed positioning shift compensation approach, the positioning error of a static target becomes 121.09m, which demonstrates the effectiveness of the proposed positioning shift compensation approach.

A three-component decomposition algorithm is proposed for polarimetric SAR data. After extracting the volume scattering component, both the orientation angle compensation and a unitary transformation are applied to the remaining matrix to derive the second and third components which are exactly consistent with either the surface scattering model or the double-bounce scattering model, respectively. We suggest calling the unitary transformation as the helix angle compensation and show that the remaining matrix is decomposed into two coherency matrices with an orientation angle difference of 45° and opposite helix angles. Experiments with real polarimetric SAR date are presented to demonstrate the effectiveness of the proposed decomposition algorithm and illustrate its several properties.

As the launch of radars, such as the Precipitation Radar (PR) on Tropical Rainfall Measuring Mission (TRMM)[1] satellite and the Surface Wave Investigation and Monitoring (SWIM) on China France Oceanography SATellite (CFOSAT)[2], that operate in low incident angles, more and more NRCS data at low incident angle will be obtained. In order to retrieve the sea surface wind speed using low incident angle NRCS, the empirical GMF of NRCS with wind speed is established. The empirical nadir reflection coefficient |R(0)2| are calculated and the empirical relationship between mean square slop s(u) and wind speed is established. The mean square slop s(u) can be retrieved by fitting the NRCS at certain incident angles with the theoretical Gaussian GMF model. Then the wind speeds are calculated using the empirical corresponding relation between mean square slop and wind speed. The retrieved wind speeds are compared with Tao and NDBC buoy. The results show that the standard deviation (STD) and bias of retrieved wind speeds are smaller than 1.7m/s and 0.1m/s respectively.

An improvement on the complete three-component model-based decomposition (C3MD) of polarimetric synthetic aperture radar (SAR) data is proposed in this letter. When analyzing the scattering mechanism of the third component in the original C3MD, an orientation angle rotation (OAR) is used. However, the deorientation is found more suitable to be applied to analyze the scattering mechanism of the third component, because the deorientation minimizes the cross-polarization power of a coherency matrix so as to make the coherency matrix much closer to that of surface scattering or double-bounce scattering. With applying the deorientation instead of the OAR used in the original C3MD, an improved C3MD algorithm is proposed. Experiments with E-SAR data are presented to illustrate the effectiveness of the improvement.

To eliminate the influence of target returns on the estimation of local sea clutter distributions, an improved iterative censoring scheme (ICS) for constant false-alarm rate detectors is proposed with two modifications. First, the proposed ICS censors out both target pixels and their four-connected neighborhood pixels from the estimation of local sea clutter distributions. Second, a novel initial detector is proposed to improve the convergence speed of ICS. The proposed initial detector, which only needs the probability of false alarms as an input parameter, is based on the sorting of all pixels under test. Experiments of ship detection with RADARSAT-2 ScanSAR wide mode images are presented to illustrate the effectiveness and improvements of the proposed ICS.

To preserve the power information in the output of polarimetric whitening filter (PWF), a modified PWF is proposed for polarimetric synthetic aperture radar (PolSAR) imagery. Besides minimizing the standard deviation to mean ratio, the modified PWF also requires its output closest to the original polarimetric vector in a minimum mean-square error sense. This principle is consistent with that of the minimum mean-square error covariance shaping. A real PolSAR image derived by SIR-C is applied for the illustration. The experimental results demonstrate that the speckle reduction performance of PWF is enhanced by the proposed modification and the power information is well preserved in the output of the modified PWF.

An improved four-component model-based target decomposition scheme for polarimetric synthetic aperture radar data is proposed in this letter. The reason for the emergence of the negative powers in the Yamaguchi decomposition has been analyzed, and three corresponding additional steps are added in the proposed scheme. First, the orientation angle compensation is applied to the coherency matrix. Second, the coherency matrix with the maximum entropy, i.e., the identity matrix is used as the volume scattering model instead of the traditional ones. Third, corresponding power constraints are appended to the scheme. Moreover, the densely vegetated areas and the residual areas are processed separately via the H/α/A classification in the proposed scheme. Finally, the polarimetric-scattering-characteristic-preserving classification is utilized to verify the improvements of the proposed scheme. To demonstrate the effectiveness of the decomposition, an Advanced Land Observing Satellite Phased-Array-type L-band Synthetic Aperture Radar polarimetric image acquired over Beijing, China, is analyzed, and the results are presented in this letter. With negative powers eliminated by the proposed scheme, improvements can be observed in the experimental results, particularly for the urban areas.

A four-component model-based decomposition for polarimetric synthetic aperture radar (SAR) images is proposed to deal with the ground objects with orientation angles around 45$^{\circ}$. In the previous decompositions, these special targets are mixed with the vegetated areas. With the deficiency of the previous decompositions analyzed, the ambiguity between two scattering mechanisms is clarified. A rotated Fresnel dihedral reflection model is introduced in the proposed algorithm, to model the scattering characteristics of these special targets. The nonnegative eigenvalue decomposition is applied to the remainder coherency matrix to prevent negative powers of the decomposed scattering mechanisms. Another advantage of the proposed decomposition is that it makes use of all the information provided by the coherency matrix, which remains unachieved in the previous model-based decompositions. Experimental Synthetic Aperture Radar (E-SAR) L-band polarimetric SAR data acquired over Oberpfaffenhofen, Germany, are analyzed in this letter. Experimental results indicate that the special ground objects have acquired correct scattering mechanisms, which verifies the effectiveness of the proposed method.

Visual interpretation of SAR images plays a critical role in remote sensing applications. For rapidly obtaining the image suitable for human observation, a new method of visualization for SAR images is proposed in this paper. The statistics of the SAR data, including the single channel data intensity and the span of the Pol-SAR data, are derived. The proposed method is a parameterized method based on the characteristic statistic of the SAR data. Using the AIRSAR data, the effectiveness of the proposed method is demonstrated by comparing to the traditional algorithm.

A modified Yamaguchi decomposition with deorientation is proposed for analyzing multilook polarimetric synthetic aperture radar (PolSAR) data. Deorientation is first applied to a coherence matrix; then, the coherence matrix is decomposed into four components by the Yamaguchi decomposition. A special kind of target is found for which the original Yamaguchi decomposition output is not appropriate. This problem is solved by applying deorientation prior to the Yamaguchi decomposition. Moreover, the deorientation procedure enhances double-bounce scattering from urban areas in the decomposition output. Comparisons of the Yamaguchi decompositions with and without deorientation are shown for PolSAR data from both airborne and spaceborne systems.

A pretest approach based on the complex Wishart distribution in polarimetric synthetic aperture radar (POLSAR) speckle filtering is proposed in this paper. The main principle is to select homogeneous pixels in a large-scale area in the filtering process, which is called pretesting. To preserve details and fine structures while despeckling, the homogeneous pixels are selected by comparing their 3 × 3 neighboring windows. A test statistic based on the complex Wishart distribution is used to decide the selection of homogeneous pixels. Speckle filtering is processed by summing up the homogeneous pixels with weights according to the values of their test statistics. To accelerate the pretest filter, we further propose a refined algorithm, which eliminates redundant operations without deteriorating the performance. We demonstrate the performance of the proposed algorithm by using both simulated and real airborne POLSAR data.

The polarimetric entropy (H) and the alpha angle (α) are two important parameters for analyzing polarimetric synthetic aperture radar data. However, for some special cases, the unstableness of the alpha angle is found; in addition, time consumption for extracting both the parameters will become quite tedious for very large images by pixelwise eigendecomposition. To overcome these shortcomings, a fast algorithm to calculate the polarimetric entropy and two new parameters are proposed in this paper. The first parameter has similar properties to the alpha angle but is stable; the second also behaves similar to the polarimetric entropy. More importantly, both new parameters can be derived very quickly. Like the unsupervisedHα classification, the image can also be classified based on the two new parameters. The NASA/JPL AIRSAR L-band data of San Francisco are applied in the experiment.

An improved three-component decomposition for polarimetric synthetic aperture radar (SAR) data is proposed in this paper. The reasons for the emergence of negative powers in the Freeman decomposition have been analyzed, and three corresponding improvements are included in the proposed method. First, the deorientation process is applied to the coherency matrix before it is decomposed into three scattering components. Then, the coherency matrix with the maximal polarimetric entropy, i.e., the unit matrix, is used as the new volume-scattering model instead of the original one adopted in the Freeman decomposition. A power constraint is also added to the proposed three-component decomposition. The E-SAR polarimetric data acquired over the Oberpfaffenhofen area in Germany are applied in the experiment. The results show that the pixels with negative powers are totally eliminated by the proposed decomposition, demonstrating the effectiveness of the new model.

An improved three-component decomposition for polarimetric SAR data is proposed in this paper. The reasons of the emergence of negative powers in the Freeman decomposition have been analyzed, and two improvements are included. Firstly, the deorientation process is applied to the coherency matrix before it is decomposed into three scattering components. Then, the coherency matrix with the maximal polarimetric entropy, i.e, the unit matrix is used as the new volume scattering model instead of the original one adopted in the Freeman decomposition. The E-SAR polarimetric data acquired over the Oberpfaffenhofen area in Germany are applied for experiment to demonstrate the effectiveness of the new model.

This letter attempts to address the problem on the emergence of negative eigenvalues in the coherency matrix after the compressing of multilook polarimetric synthetic aperture radar (SAR) data. A new nine-parameter expression for the eigenvalue decomposition of the coherency matrix is introduced, and a new compression algorithm is proposed for multilook polarimetric data with this expression. By comparing it with the NASA/Jet Propulsion Laboratory's Airborne SAR compression algorithm, the authors analyze the new algorithm's compression accuracy, signal-to-noise ratio, and the ability to preserve the data's polarimetric property. For polarimetric SAR data, it is important to preserve the polarimetric property of a target. The proposed algorithm has this ability which is illustrated by the comparison of the polarimetric signatures. Finally, the effectiveness of the proposed methods is demonstrated by using the experimental SAR data.