Ship Classification in SAR Image by Joint Feature and Classifier Selection

Lang, Haitao; Zhang, Jie; Zhang, Xi; Meng, Junmin

doi:10.1109/lgrs.2015.2506570

Cited by 82 publications

(44 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Sigmoid transform is a commonly used nonlinear transform [12][13][14]. Its definition is shown in Eq.…”

Section: Sigmoidccamentioning

confidence: 99%

A Robust Joint Estimation Method of Time Delay and Doppler Frequency Shift

Yu¹,

Niu²,

Zhang³

2018

dtcse

View full text Add to dashboard Cite

Abstract. In this paper, a novel concept, the Sigmoid transform based cyclic correlation, is proposed and a relative estimator named the Sigmoid transform based cyclic cross-ambiguity function is also derived to handle the joint estimation problem in the presence of impulsive noise and co-channel interference. Based on these definitions, a novel joint estimation method of the time delay and the Doppler frequency shift is proposed. Simulations have verified its superior performances over existing methods based on cyclic ambiguity function or generalized fractional lower-order cyclic cross-ambiguity function, especially under impulsive noise. Meanwhile, it is more applicable for practical applications since it is parameter-free.

show abstract

“…Sigmoid transform is a commonly used nonlinear transform [12][13][14]. Its definition is shown in Eq.…”

Section: Sigmoidccamentioning

confidence: 99%

A Robust Joint Estimation Method of Time Delay and Doppler Frequency Shift

Yu¹,

Niu²,

Zhang³

2018

dtcse

View full text Add to dashboard Cite

show abstract

“…Similarly, we parameterize the proposed cross-polarized coherency as Equation (8). The mathematical forms of the elements of the cross-polarized coherency are derived via integration using a probability density function p(θ) [21] with its peak at zero degrees (see Equation (9)), and any of the expected values can be derived using similar operations with Equation (10):…”

Section: Cross-polarized Scattering Modelmentioning

confidence: 99%

“…Synthetic aperture radar (SAR) can provide wide-area, all-day and nearly all-weather observation and, as such, has played an important role in maritime ship surveillance [1]. Methods of ship detection and classification based on single-polarization SAR data have been extensively studied [2][3][4][5][6][7][8]. Recent technological trends involving SAR are moving towards the utilization of polarimetric SAR (PolSAR) data in this paper, we refer to fully-polarimetric or quad-polarimetric SAR data as polarimetric SAR (PolSAR) data).…”

Section: Introductionmentioning

confidence: 99%

Four-Component Model-Based Decomposition for Ship Targets Using PolSAR Data

Lang

Tao

et al. 2017

Remote Sensing

Self Cite

View full text Add to dashboard Cite

Scattering mechanism (SM) analysis is a promising technique for ship detection and classification in polarimetric SAR (PolSAR) images. In this paper, a four-component model-based decomposition method incorporating surface, double-bounce, volume and cross-polarized components is proposed for analyzing the SMs of ships. A novel cross-polarized scattering component capable of discriminating between the HV scattering power generated by the oriented scatterers on ships from that by volume scatterers is proposed as a means to address the problem of volume scattering power overestimation. In the decomposition stage, by taking into account both the real and imaginary parts of the elements [T] HV (1, 3) and [T] HV (2, 3) of the observed coherency matrix, the proposed cross-polarized component can preserve the reflection asymmetry information completely, which is an essential property of man-made targets, such as ships. Based on the proposed decomposition method and an analysis of the different SMs between ships and sea clutter, a novel ship detection metric defined as M = lnis proposed. Experimental results conducted on RadarSat-2 quad-polarimetric data validate the proposed four-component decomposition method as being more suitable for analyzing the SMs of ship targets than the existing helix matrix-based decomposition methods. Additionally, we find that the proposed ship detection metric can effectively enhance the signal-to-clutter ratio (SCR) and improve ship detection performance.

show abstract

“…One possible way to solve this problem is to adopt feature selection strategy, and then combine the selected features. Lang et al [9] proposed a joint feature and classifier selection method, and Chen et al [10] developed a two-stage feature selection approach. Obviously, the performance of the methods can be limited by the feature set for selection, which means that, if the feature set is not friendly to classification task, we cannot get the desired effect through selection.…”

Section: Introductionmentioning

confidence: 99%

Ship Classification Based on MSHOG Feature and Task-Driven Dictionary Learning with Structured Incoherent Constraints in SAR Images

2018

View full text Add to dashboard Cite

Abstract:In this paper, we present a novel method for ship classification in synthetic aperture radar (SAR) images. The proposed method consists of feature extraction and classifier training. Inspired by SAR-HOG feature in automatic target recognition, we first design a novel feature named MSHOG by improving SAR-HOG, adapting it to ship classification, and employing manifold learning to achieve dimensionality reduction. Then, we train the classifier and dictionary jointly in task-driven dictionary learning (TDDL) framework. To further improve the performance of TDDL, we enforce structured incoherent constraints on it and develop an efficient algorithm for solving corresponding optimization problem. Extensive experiments performed on two datasets with TerraSAR-X images demonstrate that the proposed method, MSHOG feature and TDDL with structured incoherent constraints, outperforms other existing methods and achieves state-of-art performance.

show abstract

Ship Classification in SAR Image by Joint Feature and Classifier Selection

Cited by 82 publications

References 17 publications

A Robust Joint Estimation Method of Time Delay and Doppler Frequency Shift

A Robust Joint Estimation Method of Time Delay and Doppler Frequency Shift

Four-Component Model-Based Decomposition for Ship Targets Using PolSAR Data

Ship Classification Based on MSHOG Feature and Task-Driven Dictionary Learning with Structured Incoherent Constraints in SAR Images

Contact Info

Product

Resources

About