Synthetic aperture radar (SAR) is an active microwave imaging sensor with the capability of working in all-weather, all-day to provide high-resolution SAR images. Recently, images have been widely used civilian and military fields, such as ship detection. The scales different ships vary images, especially for small-scale ships, which only occupy few pixels lower contrast. Compared large-scale current detection methods are insensitive ships. Therefore, facing difficulties multi-scale A novel...

Ship target detection using large-scale synthetic aperture radar (SAR) images has important application in military and civilian fields. However, ship targets are difficult to distinguish from the surrounding background many false alarms can occur due influence of land area. False always with because most area SAR treated as clutter, considered unevenly distributing small targets. To address these issues, a method via CenterNet is proposed this article. As an anchor-free method, defines...

Compressive sensing (CS) is mainly concerned with low-coherence pairs, since the number of samples needed to recover signal proportional mutual coherence between projection matrix and sparsifying matrix. Until now, papers on CS always assume be a random In this paper, aiming at minimizing coherence, method proposed optimize This based equiangular tight frame (ETF) design because an ETF has minimum coherence. It impossible solve problem exactly complexity. Therefore, alternating minimization...

This study proposes a novel non‐negative matrix factorisation (NMF) variant L 1/2 ‐NMF after visualisation and analysis of the process target recognition via NMF for synthetic aperture radar (SAR) images. has been applied to obtain pattern feature in SAR considers intrinsic character physical meaning when automatic recognition. At base obtaining linear relationship between sample be recognised train samples, whole is detailed vividly visualised. Meanwhile, lots researches have done improve...

As a mission-critical sensor, SAR has been applied in environmental monitoring and battlefield surveillance; moreover, target recognition is one of the most important applications technology. However, practical applications, number samples available for training relatively small, so can be regarded as small sample problem. One main directions to solve problem realize data augmentation. Therefore, image augmentation method via Generative Adversarial Nets (GAN) proposed this paper. The uses...

In recent years, sparse representation-based techniques have shown great potential for pattern recognition problems. this paper, the problem of polarimetric synthetic aperture radar (PolSAR) image classification is investigated using classifiers (SRCs). We propose to take advantage both information and contextual by combining sparsity-based methods with concept superpixels. Based on feature vectors constructed stacking a variety signatures superpixel map, two strategies are considered...

The frequency of terahertz radar ranges from 0.1 THz to 10 THz, which is higher than that microwaves. Multi-modal signals, including high-resolution range profile (HRRP) and Doppler signatures, can be acquired by the system. These two kinds information are commonly used in automatic target recognition; however, dynamic gesture recognition rarely discussed regime. In this paper, a system using proposed, based on multi-modal signals. HRRP sequences signatures were first achieved echoes....

Incremental learning methods update the existing model with new knowledge when target data increase continuously. Open set recognition (OSR) algorithms provide classifiers a rejection option so that untrained type is identified. In this paper, an open incremental method introduced for automatic recognition, which able to recognize and learn unknown classes continually. The proposed method, (OSmIL), ensemble classifier it be updated only by data. For saving computational time storage source,...

Convolutional neural networks (CNNs) have been widely used in synthetic aperture radar (SAR) target recognition. Traditional CNNs suffer from expensive computation and high memory consumption, impeding their deployment real-time recognition systems of SAR sensors, as these low resources speed calculation. In this paper, a micro CNN (MCNN) for system is proposed. The proposed MCNN has only two layers, it compressed deep convolutional network (DCNN) with 18 layers by novel knowledge...

Under the framework of a supervised learning-based automatic target recognition (ATR) approach, performance is primarily dependent on amount training samples. However, shortage in samples consistent issue for ATR. In this article, we propose new image to generation method, called label-directed generative adversarial networks (LDGANs), which will provide labeled be used model training. We define an entirely loss function LDGAN, utilizes Wasserstein distance replace original measurement...

When adding new tasks/classes in an incremental learning scenario, the previous recognition capabilities trained on training data can be lost. In real-life application of automatic target (ATR), part samples may able to used. Most methods have not considered how save key samples. this article, class boundary exemplar selection-based (CBesIL) is proposed form exemplars. For selection, selection method based local geometrical and statistical information proposed. And when classes continually,...

In this paper, we propose a variational multiphase segmentation framework for synthetic aperture radar (SAR) images based on the statistical model and active contour methods. The proposed method is inspired by multiregion level set partition approaches but with two improvements. First, an energy functional which combines region information edge defined. regional term <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">G</i> <sup...

The classic ship detection methods in synthetic aperture radar (SAR) images suffer from an extreme variance of scale. Generating a set proposals before operation can effectively alleviate the multi-scale problem. In order to construct scale-independent proposal generator for SAR images, we suggest four characteristics ships and corresponding procedures this paper. Based on these procedures, put forward framework explore proposals. designed mainly contains two stages: hierarchical grouping...

Although deep learning methods have achieved great success in synthetic aperture radar automatic target recognition (SAR ATR), their accuracies decline sharply as new classes are learned, which is known catastrophic forgetting. The overlapping or confusion between the representations of and old feature space main cause In this paper, Incremental Class Anchor Clustering (ICAC) proposed to address issue. ICAC solves problem from three perspectives: first, how learn classes; second, enable...

In synthetic aperture radar (SAR) target recognition, the amount of data increases continuously, and thus SAR automatic recognition (ATR) systems are required to provide updated feature models in real time. Most recent extraction methods have use both existing new samples retrain a model every time is acquired. However, this repeated calculation leads an increased computing cost. paper, dynamic learning method called incremental nonnegative matrix factorization with L p sparse constraints (L...

In this paper, a new Region-based Convolutional Neural Networks (RCNN) method is proposed for target recognition in large scene synthetic aperture radar (SAR) images. To locate and recognize the targets SAR images, there are three steps traditional procedure: detection, discrimination, classification recognition. Each step supposed to provide optimal processing results next step, but difficult implement real-life applications because of speckle noise inefficient connection among these...

Template-matching-based approaches have been developed for many years in the field of synthetic aperture radar (SAR) automatic target recognition (ATR). However, performance template-matching-based is strongly affected by two factors: background clutter and noise size data set. To solve problems mentioned above, a multilevel reconstruction-based multitask joint sparse representation method proposed this paper. According to theory attributed scattering center (ASC) model, SAR image exhibits...

In recent years, deep convolutional neural networks (DCNNs) have been widely used in the task of ship target detection synthetic aperture radar (SAR) imagery. However, vast storage and computational cost DCNN limits its application to spaceborne or airborne onboard devices with limited resources. this paper, a set lightweight for SAR are proposed. To obtain these networks, paper designs network structure optimization algorithm based on multi-objective firefly (termed NOFA). our design, NOFA...

In this letter, we propose a target detection approach in high-resolution synthetic aperture radar (SAR) images by using the information measurement of superpixels. This study aims to transform basic cell SAR from pixel patch through superpixel algorithm. Moreover, taking advantage rich statistical character patch, an measurement, including self-information and entropy, is utilized measure difference between patches. Self-information relative value patches, while entropy used describe change...