基于对称Gabor小波滤波的舰船目标识别方法

A Study on Ship Target Recognition Based on Symmetrical Gabor Wavelet Filtering

文中构建一种对称Gabor小波滤波器(SGWF),能够对逆合成孔径雷达(ISAR)像中舰船进行识别。由Gabor小波滤波器构建的SGWF滤波器具有上下对称结构,对镜像翻转的图像滤波后仍具有镜像对称性。使用SGWF与奇异值分解(SVD)相结合来提取特征,使得该方法可以避免因多普勒变化导致的ISAR正像和倒像的检测问题。SGWF可在不同尺度和方向对图像进行滤波,充分反映图像的纹理特征和强度特征。在实验中,使用九艘民船的实测ISAR像进行识别,通过在成像过程加入高斯白噪声来检测算法对噪声的鲁棒性,并检验图像分块数量对算法性能的影响,以及检验训练样本数量对算法性能的影响,识别结果证明了算法的有效性。

In this paper, a symmetrical gabor wavelet filter（SGWF） is proposed to identify the ship target. The SGWF constructed by gabor wavelet filter（GWF） has up-down symmetry, and the symmetrical images filtered by SGWF also have symmetrical structure. SGWF and singular value decomposition （SVD） are combined to extract features from ISAR images, and it avoids detecting the ISAR image directions before recognition. The ISAR image is filtered by SGWF in arbitrary directions and scales, and SGWF can depict the texture and intensity of ISAR image sufficiently. In the experiment, the ISAR images of 9 ships are used for recognition, and the recognition results show that the proposed method is efficient and robust to Gaussian noise. Gaussian noise is added in the process of ISAR imaging to detect the noise robustness. Simultaneity, the ISAR image is divided into different block numbers and different sample number is chosen for training to research their influence on method. The experimental results demonstrate the effectiveness of proposed method.