&lt;title&gt;ATR performance using enhanced resolution SAR&lt;/title&gt;

Novak, Leslie M.; Benitz, Gerald R.; Owirka, G.J.; Bessette, Loretta A.

doi:10.1117/12.242035

Cited by 27 publications

(14 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These are undesirable image characteristics for ATR systems based on prominent peak processing [26]. The 2-D version of Capon's MVM [27], [28] is a popular choice in SAR imaging [1], [2], [3], [29], [30] and recently there has been increasing interest in its computational aspects. The interest comes from the argument that the main burden of the MVM does not come from the correlation estimate formation and inversion but from the computation of the estimator over all the needed frequencies.…”

Section: Experimental Comparison To Other Superresolution Methodsmentioning

confidence: 99%

“…Improvements in resolution of prominent scatterers, speckle reduction, enhancement of contrast, and CFAR detectability of targets in clutter are among the reported results [1]. Superresolution methods have also proven to increase significantly the performance of Automatic Target Recognition (ATR) systems [2]. A one-dimensional (1-D) nonparametric procedure to extrapolate a narrowband signal, denoted Adaptive Weighted Norm Extrapolation (AWNE) [9], has been used to enhance resolution in SAR imaging by row-column processing [7].…”

Section: Introductionmentioning

confidence: 94%

“…However, objective comparison metrics cannot be used with these full SR images, as in previous work A. E. BRITO, S. H. CHAN AND S. D. CABRERA reported in [7], [8] because of the lack of a ground truth image to compare with. An alternative approach is the objective evaluation of the full SR images based on their impact on the performance of an ATR system [2], [26]. Also, general image quality metrics that allow a comparison among various superresolution generated images coming from different methods [1], [3], [4], [5], [6] can be defined and used.…”

Section: Target Zooming Schemementioning

confidence: 99%

“…There is an increasing interest in the use of superresolution methods to form Synthetic Aperture Radar (SAR) images [1], [2], [3], [4], [5], [6], [7], [8]. The motivation for applying these methods to SAR, as discussed by DeGraaf [1], is to improve resolution, to suppress sidelobe artifacts, and to reduce speckle in comparison with traditional Fourier imaging techniques.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

SAR Image Superresolution via 2-D Adaptive Extrapolation

Brito¹,

Chan²,

Cabrera³

2003

Radar Signal Processing and Its Applications

View full text Add to dashboard Cite

Abstract. In this paper, we present a description of a nonparametric two dimensional (2-D) procedure to extrapolate a signal, an extension of the Adaptive Weighted Norm Extrapolation (AWNE) method, and illustrate its application to SAR image formation. The benefits of the AWNE procedure are shown for synthetic data and for MSTAR data. Once the phase history is recovered, the AWNE method is applied to a subaperture or to the full set of frequency samples to extrapolate them to a larger aperture from which a superresolved complex SAR image is obtained. Use of the 2-D AWNE procedure proves to be superior to its one-dimensional separable version by reducing undesirable effects such as sidelobe interference, and variability in energy of the extrapolated data from row to row and column to column. To assess the performance of AWNE in enhancing prominent scatterers, reducing speckle, and suppressing clutter, we compare the superresolved images to the images formed with the traditional Fourier technique starting from the same phase history data. Fourier images are also compared with superresolved images formed using less data in order to assess the quality of the extrapolation and to quantify the method's ability to recover lost resolution. We illustrate performance by visual comparison and by the use of a geometric constellation of prominent point scatterers of the targets extracted from the images. A brief comparison with the 2-D versions of Capon and the linear prediction methods is illustrated and a hybrid AWNE/Capon approach is proposed.

show abstract

Section: Experimental Comparison To Other Superresolution Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 94%

Section: Target Zooming Schemementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

SAR Image Superresolution via 2-D Adaptive Extrapolation

Brito¹,

Chan²,

Cabrera³

2003

Radar Signal Processing and Its Applications

View full text Add to dashboard Cite

show abstract

“…The process of SAR ATR can be summarized as finding regions of interest (ROIs) in the observed SAR image and classifying the category of each ROI (e.g., T72 or BTR70) [4]. Some earlier methods of SAR ATR can be found in [5][6][7][8][9].…”

Section: Introductionmentioning

confidence: 99%

D-ATR for SAR Images Based on Deep Neural Networks

et al. 2019

View full text Add to dashboard Cite

Automatic target recognition (ATR) can obtain important information for target surveillance from Synthetic Aperture Radar (SAR) images. Thus, a direct automatic target recognition (D-ATR) method, based on a deep neural network (DNN), is proposed in this paper. To recognize targets in large-scene SAR images, the traditional methods of SAR ATR are comprised of four major steps: detection, discrimination, feature extraction, and classification. However, the recognition performance is sensitive to each step, as the processing result from each step will affect the following step. Meanwhile, these processes are independent, which means that there is still room for processing speed improvement. The proposed D-ATR method can integrate these steps as a whole system and directly recognize targets in large-scene SAR images, by encapsulating all of the computation in a single deep convolutional neural network (DCNN). Before the DCNN, a fast sliding method is proposed to partition the large image into sub-images, to avoid information loss when resizing the input images, and to avoid the target being divided into several parts. After the DCNN, non-maximum suppression between sub-images (NMSS) is performed on the results of the sub-images, to obtain an accurate result of the large-scene SAR image. Experiments on the MSTAR dataset and large-scene SAR images (with resolution 1478 × 1784) show that the proposed method can obtain a high accuracy and fast processing speed, and out-performs other methods, such as CFAR+SVM, Region-based CNN, and YOLOv2.

show abstract