基于Gabor幅值特征和相位特征相融合的ISAR像目标识别

摘要

A new Inverse Synthetic Aperture Radar (ISAR) target recognition method with the fusion of Gabor magnitude and phase feature is proposed. Firstly, the corresponding Gabor Magnitude Maps (GMMs) and Gabor phase information are obtained by convolving the ISAR image with multi-scale and multi-orientation Gabor filters. Secondly, each GMM is divided into several non-overlapping rectangular units, and the histogram of unit is computed and combined as the magnitude histogram feature. Thirdly, the local Gabor phase pattern is obtained by combining quadrant bit coding with local XOR pattern, and the block histogram feature is extracted from the local Gabor phase pattern. Then, the fusion of the Gabor magnitude and phase feature is used as the feature of ISAR image. Finally, five-type aircraft models are classified by using a nearest neighbor classifier with 2c as a dissimilarity measure in the computed feature space. The recognition method is tested on ISAR data simulated from Greco electromagnetic soft ware. Compared with other recognition methods, the numerical results show that the proposed method is effective and has higher recognition performance.%　　该文提出一种基于 Gabor 小波变换幅值特征和相位特征相融合的 ISAR 像目标识别算法。首先将 ISAR像进行Gabor多尺度分析，对不同尺度、不同方向的Gabor幅值图像划分为若干矩形不重叠的子块，分别计算每个子块的直方图分布，将其联合起来作为Gabor的幅值特征；然后结合象限二进制编码和局部异或算子提取Gabor相位信息的局部相位模式，对得到的局部相位模式同样提取分块直方图特征；再把提取的幅值特征和相位特征相融合作为ISAR像最终的Gabor特征；最后在2c统计量作为不相似度量计算的特征空间里，采用最近邻分类器完成5类目标的分类识别。通过使用Greco电磁软件仿真的5类目标的ISAR数据对该方法进行目标识别的验证，并与目前已有的几种Gabor特征提取方法进行了对比，结果表明，该方法是可行且有效的，能够明显地提高识别率。