A Multi-Scale Local Phase Quantization Plus Biomimetic Pattern Recognition Method for Sar Automatic Target Recognition

Abstract

Abstract-Synthetic aperture radar (SAR) automatic target recognition (ATR) has been receiving more and more attention in the past two decades. But the problem of how to overcome SAR target ambiguities and azimuth angle variations has still left unsolved. In this paper, a multi-scale local phase quantization plus biomimetic pattern recognition (BPR) method is presented to solve these two difficulties. By applying multiple scales local phase quantization (LPQ) on the observed SAR images, the blur and azimuth invariant features can be extracted, and these features are fusion at consecutive multiple scales to achieve better results. Then PCA method is applied to further reduce the feature dimension and achieve its efficiency. Finally, high dimensional space geometry covering method based on BPR theory is adopted to construct hyper sausage neuron links for target recognition. Experiments on the MSTAR database show that the proposed method can achieve satisfying recognition accuracy compared with other stateof-the-art methods.

Imaging with Electrons, X-rays, and Microwaves

“…From a considerable body of research, I will quote just two papers that together give a reasonably up-to-date representation of the present state of urban remote sensing. The first deals with real-time SAR simulation and pattern recognition in urban areas (Balz, Becker, Haala, & Martin, 2008), and the second takes a systematic look at automatic pattern recognition in SAR images for military and commercial applications (Zhai, Li, Gan, & Ying, 2013 The geometric theory of diffraction (GTD) and its uniform versions uniform theory of diffraction (UTD); Kouyoumjiam & Pathak (1974) and uniform asymptotic theory (UAT); Lee & Deschamps (1976) have been used with much success for a wide variety of high-frequency diffraction problems. However, the range of problems to which GTD can be applied is limited by the availability of the solutions to the canonical problems from which the diffraction coefficients of GTD are calculated.…”

Sliding Scattering Center Model for Extended Streamlined Targets

“…Different models were proposed for describing the features of scattered waves by the complex electromagnetic environments [1][2][3][4][5]. For extended target in free space, scattering center model [7][8][9][10][11][12] is the most effective model for describing the features of scattered waves by the target. For higher accuracy of radar signal simulation or parameter estimation, the scattering center model should be built based on the scattering characteristics of the extended target rather than the signatures shown in radar or optical images on the consideration of target as a set of fixed scattering centers.…”

“…From the view of radar signal processing, scattering center model is more practical than scattering characteristics as it is provides straight relationship between the signatures in radar images and the physical features of targets, and thus are broadly used in many radar applications, such as shape, velocity and other physical parameters estimation [6,7], automatic target recognition (ATR) [8,9], radar image interpretation [10][11][12], and radar data compression [13][14][15], etc.. The scattering centers due to scattering sources at discontinuities of surface, such as spires, corners and gaps, etc., have been of concern in applications of geometry parameter estimation, radar target tracking and recognition for decades [16][17][18], for their scattering characteristics being stable within a relatively wide radar observation angle.…”

Cfar Target Detection in Ground Sar Image Based on Kk Distribution

“…It is reported that typical sensor platforms is able to provide wide area search coverage (approximately 100 km 2 area per minute) at 1.0 m × 1.0 m resolution [6]. This attractive capability has opened new interesting research fields in both civil and military contexts, such as target detection [7,8], target recognition [9,10], navigation [11], etc.. Target detection is one of the core applications of SAR remote sensing and has received considerable attentions in the recent years, such as CFAR method [12][13][14][15][16][17][18], the GLRT method [19], the extended fractal-based method [20], the wavelet transform-based method [21], etc.. Among these kinds of methods, the CFAR method has been most widely used for it is capable of maintaining the constant false alarm probability at a certain level in non-stationary background through an adaptive threshold.…”