基于特征配准的ISAR图像方位定标方法

逆合成孔径雷达(ISAR)成像利用目标相对雷达视线的姿态变化形成的合成孔径获得方位高分辨，成像方位为多普勒轴，通常需要估计目标的有效转动速度以实现ISAR图像的方位定标从而体现目标真实尺寸。现有算法通常利用信号的运动参数估计和图像整体配准。该文提出利用子孔径ISAR图像的特征提取和配对，根据特征点坐标估计目标的有效转角速度。首先利用尺度不变特征变换(SIFT)和快速鲁棒特征(SURF)对两幅ISAR图像进行特征点提取；然后分别采用最短欧氏距离和随机采样一致性(RANSAC)进行特征点的匹配和失配点的剔除；最后根据配对特征点的坐标和能量估算有效转角速度，实现ISAR图像方位定标。仿真数据和实测数据验证了该算法的精确性和稳健性。

A Novel Cross-range Scaling Algorithm for ISAR Images Based on Feature Registration

As Inverse SAR (ISAR) imaging utilizes synthetic aperture with aspects changes related to the Radar Line of Sight (RLOS) to acquire azimuth resolution, the accurate estimation of rotated velocity is pivotal for the geometric scaling of ISAR images to measure the real size of a target. Compared with current methods by estimating motional parameters and the integrated images registration, this paper proposes a novel algorithm by extracting and registering the interested points of ISAR images from sub-aperture data, which provides the points coordinate-locations to calculate the virtual rotated velocity. First, adequate interested points are extracted from two sub-aperture images by Scale Invariant Feature Transform (SIFT) and Speeded-Up Robust Features (SURF). Those points are then pinpointed by matching and re-matching with the minimum Euclid- distance and RANdom SAmple Consensus (RANSAC) principles, respectively. Finally, the rotated velocity, a premise to acquire the cross-resolution, can be estimated to achieve the precise target scaling. Simulated and real data validate the effectiveness and robustness of the proposed algorithm.