Bin Lei

Transfer Learning with Deep Convolutional Neural Network for SAR Target Classification with Limited Labeled Data

Tremendous progress has been made in object recognition with deep convolutional neural networks (CNNs), thanks to the availability of large-scale annotated dataset. With the ability of learning highly hierarchical image feature extractors, deep CNNs are also expected to solve the Synthetic Aperture Radar (SAR) target classification problems. However, the limited labeled SAR target data becomes a handicap to train a deep CNN. To solve this problem, we propose a transfer learning based method, making knowledge learned from sufficient unlabeled SAR scene images transferrable to labeled SAR target data. We design an assembled CNN architecture consisting of a classification pathway and a reconstruction pathway, together with a feedback bypass additionally. Instead of training a deep network with limited dataset from scratch, a large number of unlabeled SAR scene images are used to train the reconstruction pathway with stacked convolutional auto-encoders (SCAE) at first. Then, these pre-trained convolutional layers are reused to transfer knowledge to SAR target classification tasks, with feedback bypass introducing the reconstruction loss simultaneously. The experimental results demonstrate that transfer learning leads to a better performance in the case of scarce labeled training data and the additional feedback bypass with reconstruction loss helps to boost the capability of classification pathway.

Fast Vessel Detection in Gaofen-3 SAR Images with Ultrafine Strip-Map Mode

This study aims to detect vessels with lengths ranging from about 70 to 300 m, in Gaofen-3 (GF-3) SAR images with ultrafine strip-map (UFS) mode as fast as possible. Based on the analysis of the characteristics of vessels in GF-3 SAR imagery, an effective vessel detection method is proposed in this paper. Firstly, the iterative constant false alarm rate (CFAR) method is employed to detect the potential ship pixels. Secondly, the mean-shift operation is applied on each potential ship pixel to identify the candidate target region. During the mean-shift process, we maintain a selection matrix recording which pixels can be taken, and these pixels are called as the valid points of the candidate target. The l1 norm regression is used to extract the principal axis and detect the valid points. Finally, two kinds of false alarms, the bright line and the azimuth ambiguity, are removed by comparing the valid area of the candidate target with a pre-defined value and computing the displacement between the true target and the corresponding replicas respectively. Experimental results on three GF-3 SAR images with UFS mode demonstrate the effectiveness and efficiency of the proposed method.

Airplane Recognition in TerraSAR-X Images via Scatter Cluster Extraction and Reweighted Sparse Representation

Target recognition in synthetic aperture radar (SAR) images has become a hotspot in recent years. The backscattering characteristic of target is a significant issue taken into consideration in SAR applications. Almost all of the previous work focus on the scatter point extraction to depict the backscattering characteristic of the target; however, a point-target corresponds to a region rather than a single point due to the convolution during the imaging. Based on this fact, we first analyze the extent to how a point-target spreads, then propose a novel scatter cluster extraction (SCE) method, and utilize the scatter cluster as the feature to solve the airplane recognition problem in SAR images. In practice, there often exist interfering objects near the target to be classified. To overcome this issue, we design a reweighted sparse representation (RSR)-based automatic purifying method by assigning a weight to each element of the feature iteratively according to the representation error. Since the element with large representation error always corresponds to the interfering objects, we give it a small weight, consequently suppressing the influence of the interference. Experimental results demonstrate that the proposed SCE method outperforms the traditional scatter point extraction-based method as well as some state-of-the-art methods. The comparison result also validates the effectiveness of the proposed RSR method.

Synthetic Aperture Radar Image Synthesis by Using Generative Adversarial Nets

Synthetic aperture radar (SAR) image simulators based on computer-aided drawing models play an important role in SAR applications, such as automatic target recognition and image interpretation. However, the accuracy of such simulators is due to geometric error and simplification in the electromagnetic calculation. In this letter, an end-to-end model was developed that could directly synthesize the desired images from the known image database. The model was based on generative adversarial nets (GANs), and its feasibility was validated by comparisons with real images and ray-tracing results. As a further step, the samples were synthesized at angles outside of the data set. However, the training process of GAN models was difficult, especially for SAR images which are usually affected by noise interference. The major failure modes were analyzed in experiments, and a clutter normalization method was proposed to ameliorate them. The results showed that the method improved the speed of convergence up to 10 times. The quality of the synthesized images was also improved.

The GF-3 SAR Data Processor

The Gaofen-3 (GF-3) data processor was developed as a workstation-based GF-3 synthetic aperture radar (SAR) data processing system. The processor consists of two vital subsystems of the GF-3 ground segment, which are referred to as data ingesting subsystem (DIS) and product generation subsystem (PGS). The primary purpose of DIS is to record and catalogue GF-3 raw data with a transferring format, and PGS is to produce slant range or geocoded imagery from the signal data. This paper presents a brief introduction of the GF-3 data processor, including descriptions of the system architecture, the processing algorithms and its output format.

Unsupervised Mixture-Eliminating Estimation of Equivalent Number of Looks for PolSAR Data

This paper addresses the impact of mixtures between classes on equivalent number of looks (ENL) estimation. We propose an unsupervised ENL estimator for polarimetric synthetic aperture radar (PolSAR) data, which is based on small sample estimates but incorporates a mixture-eliminating (ME) procedure to automatically assess the uniformity of the estimation windows. A statistical feature derived from a combination of linear and logarithmic moments is investigated and adopted in the procedure, as it has different mean values for samples from uniform and nonuniform windows. We introduce an approach to extract the approximated sampling distribution of this test statistic for uniform windows. Then the detection is conducted by a hypothesis test with adaptive thresholds determined by a nonuniformity ratio. Finally the experiments are performed on both simulated and real SAR data. The capability of the unsupervised ME procedure is verified with simulated data. In the real data experiments, the ENL estimates of Flevoland and San Francisco PolSAR images are analyzed, which show the robustness of the proposed ENL estimation for SAR scenes with different complexities.

A fast automatic U-distribution segmentation algorithm for polsar images

A novel unsupervised, non-Guassian and contextual clustering algorithm for segmentation of polarimetric SAR images has been presented in [1]. This represents one of the most advanced PolSAR unsupervised statistical segmentation algorithm and uses the doubly flexible, two parameter, U-distribution model for the PolSAR statistics. However complexity of the probability density function leads to high time consumption. This paper investigate the key dependent variable in the U-distribution model and find a new parameter domain where the PDFs are smooth. Then a one-dimensional look-up table is set in this domain with nodes number determined by corresponding Fourier spectrum and is adopted to avoid re-evaluating the numerical integral in PDF to calculate class posteriori probabilities for every sample. The proposed strategy is incorporated in the standard segmentation algorithm. Prototype test has been carried out to validate the effectiveness of the proposed method.

An Improved Shape Contexts Based Ship Classification in SAR Images

In synthetic aperture radar (SAR) imagery, relating to maritime surveillance studies, the ship has always been the main focus of study. In this letter, a method of ship classification in SAR images is proposed to enhance classification accuracy. In the proposed method, to fully exploit the distinguishing characters of the ship targets, both topology and intensity of the scattering points of the ship are considered. The results of testing the proposed method on a data set of three types of ships, collected via a space-borne SAR sensor designed by the Institute of Electronics, Chinese Academy of Sciences (IECAS), establish that the proposed method is superior to several existing methods, including the original shape contexts method, traditional invariant moments and the recent approach.

Accurate sea–land segmentation using ratio of average constrained graph cut for polarimetric synthetic aperture radar data

Abstract. Separating sea surface and land areas in synthetic aperture radar (SAR) images is challenging yet of great importance to coastline extraction and subsequent coastal classification. Results of the previous state-of-art methods often suffer from a number of limitations that arise from the presence of the speckle effect and the inadequate returned signal around the boundaries. We propose a graph cut (GC)-based approach to tackle these limitations and achieve accurate sea–land segmentation results. To be more specific, as the first step, three powerful multipolarization features are extracted from the polarimetric SAR data as descriptors to fully characterize the sea area and land area. Starting from that, seeds of the sea and land are selected automatically to build the prior model for GC. Based on the prior model, we construct the undirected graph in GC using the multipolarization descriptors. Finally, we incorporate the ratio of average operator to eliminate the speckle effect and get finer results for some finer structures. Experiments on Radarsat-2 quad-polarization images demonstrate significantly improved results of our proposed algorithms compared with several state-of-the-art methods in terms of both quantitative and visual performance.

A Preliminary Analysis of Wind Retrieval, Based on GF-3 Wave Mode Data

This paper presents an analysis of measurements of the normalized radar cross-(NRCS) in Wave Mode for Chinese C-band Gaofen-3(GF-3) synthetic aperture radar (SAR). Based on 2779 images from GF-3 quad-polarization SAR in Wave Mode and collocated wind vectors from ERA-Interim, this experiment verifies the feasibility of using ocean surface wind fields and VV-polarized NRCS to perform normalized calibration. The method uses well-validated empirical C-band geophysical model function (CMOD4) to estimate the calibration constant for each beam. In addition, the relationship between cross-pol NRCS and wind vectors is discussed. The cross-pol NRCS increases linearly with wind speed and it is obviously modulated by the wind direction when the wind speed is greater than 8 m/s. Furthermore, the properties of the polarization ratio, denoted PR, are also investigated. The PR is dependent on incidence angle and azimuth angle. Two empirical models of the PR are fitted, one as a function of incidence angle only, the other with additional dependence on azimuth angle. Assessments show that the σVV0 retrieved from new PR models as well as σHH0 is in good agreement with σVV0 extracted from SAR images directly.