Sar Target Classification Using Bayesian Compressive Sensing With Scattering Centers Features

Zhang, Xinzheng; Qin, Jianhong; Li, Guojun

doi:10.2528/pier12120705

Cited by 39 publications

(30 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the development of high-resolution SAR in recent years, new progress has been made to depict SAR images with more details. Carried in the magnitude of the radar backscatter, the scattering centers are considered as distinctive characteristics of SAR target images [15]. As depicted in [10], instead of using the scatter point extraction to describe the backscattering characteristic, a scatter cluster extraction method is proposed.…”

Section: Introductionmentioning

confidence: 99%

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

2017

View full text Add to dashboard Cite

Abstract: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.

show abstract

Section: Introductionmentioning

confidence: 99%

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

2017

View full text Add to dashboard Cite

show abstract

“…By using the sparse signal recovery algorithm, the imaging quality can be improved. This has been successfully shown in SAR imaging [7,8], ISAR imaging [9,10], MIMO radar imaging [11,12].…”

Section: Introductionmentioning

confidence: 99%

Isar Imaging Based on Iterative Reweighted Lp Block Sparse Reconstruction Algorithm

Feng¹,

Zhang²

2016

PIER M

View full text Add to dashboard Cite

Abstract-Sparse signal recovery algorithms can be used to improve radar imaging quality by using the sparse property of strong scatterers. Traditional sparse inverse synthetic aperture radar (ISAR) imaging algorithms mainly consider the recovery of sparse scatterers. However, the scatterers of an ISAR target usually exhibit block or group sparse structure. By utilizing the inherent block sparse structure of ISAR target images, an iterative reweighted l p (0 < p ≤ 1) block sparse signal recovery algorithm is proposed to enhance imaging quality in this paper. Firstly, an ISAR imaging signal model is established with the aid of sparse basis, and the imaging is mathematically converted into block reweighted cost function optimization problem. Then, an iterative algorithm is used to solve the reweighted function minimization problem. In each iteration, the weights are updated based on the closed form solution of the previous iteration. The proposed method is effective to exploit the underlying block sparse structures which does not need the prior knowledge of the number of the blocks. Real data ISAR imaging results are provided to verify that the proposed algorithm in this paper can achieve better images than the images obtained by several popular sparse signal recovery algorithms.

show abstract

“…It should be noted that the scattering coefficients not at the discrete points can be interpolated from the adjacent grid points. The global scattering center model is very useful in many radar applications [12][13][14][15][16][17][18][19] as it has some good characters: 1) the global scattering center model is not particular to radar system parameters such as operating frequency, bandwidth, and waveform; 2) the target echoes can be simulated instantly based on the model, which means that the range profiles and the SAR/ISAR images can be easily reconstructed in real time; 3) the global scattering center model has explicit physical interpretation, so it can incorporate with other information. For example, when the target configuration is changed, the corresponding scattering centers can be modified to reflect the change without rebuilding a new model.…”

Section: Problem Formulationmentioning

confidence: 99%

Three-Dimensional Scattering Centers Extraction of Radar Targets Using High Resolution Techniques

Zhang¹,

Hu²,

Gao³

et al. 2014

PIER M

View full text Add to dashboard Cite

Abstract-In optical region, the scattering center model is very useful in scattering analysis, target recognition and data compression. The method based on Hough transformation performs well in most cases. However, the algorithm extracts the scattering centers one by one via a clean method, which is time consuming. To solve this problem, a novel method is proposed in this paper to extract the scattering centers. By searching the estimated 1D scattering centers, the candidate positions for 3D scattering centers are extracted. Then the candidates are discriminated by a clustering based procedure. By employing the new algorithm, the 3D scattering centers can be extracted simply and the clean step is unnecessary, which makes the procedure efficient. The experiment results of the high-frequency-electromagnetic data demonstrate the performance of the proposed method.

show abstract

Sar Target Classification Using Bayesian Compressive Sensing With Scattering Centers Features

Cited by 39 publications

References 37 publications

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

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

Isar Imaging Based on Iterative Reweighted Lp Block Sparse Reconstruction Algorithm

Three-Dimensional Scattering Centers Extraction of Radar Targets Using High Resolution Techniques

Contact Info

Product

Resources

About