<title>Multistage algorithm for detection of targets in SAR image data</title>

Abstract

A popular approach to SAR ATR is to use a hierarchical arrangement of stages, aimed at narrowing the field of interest and concentrating detection resources in those areas with the highest probability of containing a target. CFAR detectors have played a prominent role in such systems. So have morphological processing techniques, but to a lesser extent. The work reported in this paper follows the traditional hierarchical approach and employs CFAR and morphological principles. With the goal of improving detection accuracy and reducing the false alarm rate, we have explored and developed some new variations on the theme.The baseline CFAR and morphology algorithms have been significantly modified in proposed techniques to better match SAR target signatures. In addition, the modified algorithms have the ability to locate targets in a fast and efficient manner, allowing for the possibility of real-time implementation. Experimental evaluations indicate that both approaches perform well, with further performance gains noted when the algorithms are merged. The advantages and drawbacks of each algorithm are presented and experimental results are shown for a variety of test cases.

Shape-based recognition of targets in synthetic aperture radar images using elliptical Fourier descriptors

“…The SVM classification schemes were implemented using a freely available package for Matlab™ called the Support Vector Machine Toolbox Version 0.5 23 . In general, it was difficult to compare the EFD-based classification results with the ones of previous studies, as many of them used subsets of the available images in the MSTAR dataset for both 3-target and 10-target ATR problems 11,12,24 . Some studies considered the pose (azimuth) to be known a priori, which significantly simplified the ATR problem (see for example Cetin 9 and others 11 ) and only a few did otherwise 3,5,10,25 .…”

Target detection based on multiresolution fractal analysis

“…Moreover, fast techniques for speeding up further the WF feature extraction process will be investigated. In this work, the feature extraction process consists of three steps: (a) a gradient computation step, which involves two filtering operations and a combination of the filter outputs (equation (9)), (b) a partition function computation step, which involves one max/one min operations in an annular window, and one moving average, for two different scales (equation (10), and (c) a slope calculation step (equation (11)). …”

“…In order to circumvent the disadvantages of intensity-based techniques, texture-based techniques [9][10][11][12][13][14] have been investigated for target detection in SAR imagery. Texture-based techniques assume that SAR images can be represented…”

Fast detecting and locating groups of targets in high-resolution SAR images

“…To realize the necessary processing speed up, the detection algorithm must exhibit low computational complexity and a reasonable false alarm rate. For the algorithms of target detection in SAR images, the most widely accepted is the two-parameter CFAR detector [14][15][16][17] which is under the condition that the background clutter is Gaussian. This two-parameter CFAR detector consists of three regions: the target area, the guard area and the background clutter area, as shown in Fig.…”