一种基于盲同态解卷积的SAR成像自聚焦算法

An image autofocus algorithm using blind homomorphic deconvolution for synthetic aperture radar

针对大气扰动及飞行平台不稳引起机载合成孔径雷达(SAR)图像散焦的问题,提出了一种新的自聚焦算法来估算运动误差。该算法是基于运动误差函数及场景散射函数平滑特性的差异来进行估计的,利用复对数变换将图像的幅度与相位信息分离,进而分别对相位及幅度信息进行处理。运动误差通常为慢变函数,而场景的散射信息具有某种随机特性。因此,经过复对数变换后,运动误差及散射信息可以通过滤波器进行分离,将相位中的随机噪声去除,从而保留了慢变的运动误差函数。为了去除噪声信息,需要建立一个平滑滤波器,利用Daubechies小波的尺度函数构造Riesz基向量,从而建立了正交子空间,通过所建立信号子空间及噪声子空间组建平滑滤波器,最终可以获得准确的运动误差。在实验部分,分别利用仿真数据及实测数据对本文方法进行验证,最终结果分析表明该方法具有很高的估计精度及执行效率。

Synthetic aperture radar （SAR） image suffers from the deterioration due to the unknown phase error caused by unstable platform and atmosphere perturbation. A novel autofocus algorithm is presented in this paper to obtain phase error. The proposed algorithm is put forward based on the differences of smoothness properties for log-spectrum of motion error and image reflectivity. The differences could be employed to discriminate the motion error and reflectivity function. The tog- spectrum of motion error is a slow-varying function, while the log-spectrum of image reflectivity owns some statistical properties which is similar to jagged function distribution. Then, motion error can be separated by applying a proper smoothing filter to the log-spectrum of blurred image. We set up a model that the log-amplitude spectrum and phase of spectrum for blurred image are processed through different smoothing filter functions. The log-spectrum of amplitude is recovered by cur- rent de-noising algorithms and phase is restored through phase-unwrapping and smoothing filter. It is demonstrated that the amplitude and phase of motion error can be reliably restored from the blurred SAR image and refiectivity function of image can be accurately reconstructed. In this paper, Riesz basis is constructed by scaling function of Daubechies wavelet function. An orthogonal subspace is built. Finally, a smoothing filter is applied to the derivative of compressed data. Then, motion er- ror can be obtained. In order to demonstrate the performance of the proposed method, simulation data and real data are pro- cessed to verify the proposed algorithm. The analysis illustrates that this algorithm could obtain accurate motion error and possesses higher implementation efficiency.