传统合成孔径雷达(SAR)图像在高频率、空间尺度等条件的影响下,存在图像特征校准率低、相位一致性不具备以及在非线性条件下稳定性差的不足。为了解决这一问题,文中提出了基于相位一致和稳定非线性SIFT 算法的SAR 图像校准,很好地解决了原始SAR 图像下校准率低的问题。首先,采用尺度不变特征转换(SIFT)算法,对SAR 图像进行了处理;其次,进行了无用特征点的剔除,使得检测到的图像基本达到了粗配准的目的,同时对图像进行了欧氏距离匹配,使图像特征点满足匹配。实验结果表明,文中算法满足在SAR 图像下的校验,并通过对比相关数据表明算法满足相位一致与非线性条件下的稳定性。对比其他算法,该算法已基本具备SAR 图像高效校准的要求,较改进前稳定性、可靠性等均有一定的提升。

Under the influence of conditions such as high frequency and spatial scale, traditional synthetic aperture radar (SAR) images have the disadvantages of low accuracy of image features, lack of phase consistency, and poor stability under non-linear conditions. In order to solve this problem, SAR image calibration based on the phase-consistent and stable nonlinear SIFT algorithm is proposed, which solves the problem of low accuracy in the original SAR image. Firstly, SIFT algorithm is used to process the SAR image; Secondly, the useless feature points are eliminated, so that the detected image basically achieves the purpose of coarse registration. At the same time, the image is matched with Euclidean distance to make the image feature points meet the matching. The experimental results show that the algorithm meets the verification in SAR image, and the comparison of related data shows that the algorithm meets the stability under the conditions of phase consistency and nonlinearity. Compared with other algorithms, the algorithm basically meets the requirements for high-efficiency calibration of SAR images, and its stability and reliability have been certain improved.

TN958