基于伪魏格纳分布分解的合成孔径雷达图像目标与阴影分割

Target and Target Shadow Segmentation of Synthetic Aperture Radar Image Based on the Pseudo Wigner distribution Decomposition

在二维魏格纳分布的框架内,针对魏格纳变换的交叉项问题和计算量大的问题,提出了合成孔径雷达图像局部伪魏格纳变换的目标和目标阴影的分割方法.首先,将合成孔径雷达图像进行二维伪魏格纳变换,得到各像素点的二维能量谱图;然后提取各像素点的二维能量谱图对应位置值形成多个不同频段的与原图像同大小的能量谱图;最后,对不同频段的能量谱图采用不同的处理方法后,将各能量谱图相加处理后形成区域标识图像,最终得到原图像的目标和目标阴影分割图像.本文利用该方法对MSTAR切片图像进行了分割试验,并对分割图像与频谱最大值距离或方位分割算法和基于双参量CFAR与隐马尔科夫联合分割算法进行了分割图像对比度对比.实验结果表明,采用本文算法的合成孔径雷达分割图像,对比度明显提高,且保留了目标图像细节.

In order to sovle the problem of expensive cost and the interference of cross terms in two-dimensional（2D） Wigner-Ville distribution（WVD） category,a novel target and target shadow segmentation method of synthetic aperture radar（SAR） images based on the two-dimensional Pseudo Wigner-Ville distribution（PWVD） decomposition were proposed.Firstly,2D local energy maps of every pixel of original SAR image were obtained through local 2D-PWVD transform to SAR image.Then,the resultant data which are 2D local energy maps of every pixel of original SAR image could be arranged as N energy maps which are the same size with original SAR image.Finally,the different processing methods to the resultant data in frequency dimension were adopted according to different energy maps.The marker region image of target and target shadow was formed through adding the filtered energy maps.The segmentation experiment for MSTAR chips were taken using the proposed method.The comparisons between segmentation images of our method and segmentation images of the algorithm based on distance or azimuth of max energy,and between segmentation images of our method and segmentation images of the compounding segmentation algorithm based on two-parameter CFAR and Markov Random Field were carried out.The results show that the segmentation images by the presented algorithm preserve the abundant detail information of original images.