目标检测是自动目标识别(ATR)的第一个阶段。研究合成孔径雷达(SAR)图像目标检测问题,提出了一种基于Rayleigh分布的CFAR快速检测算法,将CFAR检测分成水平和垂直CFAR检测两步进行。利用相邻点参考窗口的重合及图像的分布特性,提高了参数估计的效率。算法同时利用目标方差特性以减少虚警率。对MSTAR数据进行实验,结果表明该算法具有较好的性能。

Target detection is the first stage of automatic target recognition(ATR). A rapid Constant False Alarm Rate (CFAR) detection algorithm is proposed in this paper. This detection algorithm is based on Rayleigh clutter distribution assumption and is composed of two stages: a horizontal CFAR and a vertical CFAR. The superposition of the reference windows of adjacent pixels is used to speed up the estimation of clutter mean, and the characteristic of clutter distribution is used to speed up the estimation of clutter variance. The standard variance characteristic of target data is used to reduce false alarm rate. A weighted counter filter is used to eliminate the sparse bright pixels, and the connected regions whose area is within a certain bound are detected as region of interest. Experiment using the MSTAR pubic data set shows the power of this approach in SAR target detection.