基于相位差异法的多帧遥感降质图像复原仿真

Simulation of Multi-Frame Remote Sensing Degraded Image Restoration Based on Phase Difference Method

为降低大气湍流和光学系统对成像效果的干扰,提高画面可用信息质量,提出基于相位差异法的多帧遥感图像复原方法。构建多帧遥感图像模型,利用低通滤波器去除明显噪声,避免振铃现象发生。采用相位差异法反演遥感成像流程,凭借一对处于离焦面和焦面位置的CCD相机,连续采集短曝光图像,得到目标瞬时波前的相位数据。将数据代入极坐标下的泽尼克式中,获得焦面和离焦面的光学传递函数和光场分布,利用正则化约束复原目标函数和差异法代价函数,求解似然函数的最小值,即实现遥感降质图像复原。实验结果表明,所提方法复原后图像的灰度平均梯度、拉普拉斯梯度模、Fisher信息量均大于原始图像值,证明经过复原后图像的细节特征更丰富,分辨率更高,复原质量好。

In order to reduce the interference of atmospheric turbulence and optical system on the imaging effect and improve the quality of available information,this article puts forward a method of restoring multi-frame remote sensing images based on phase difference.Firstly,a model of multi-frame remote sensing image was built,and then a low-pass filter was used to remove obvious noise,thus avoiding the ringing phenomenon.Secondly,the phase difference method was adopted to invert the imaging process.Thirdly,a pair of CCD cameras located at the defocus plane and the focal plane was used to collect short-exposure images continuously and thus to obtain the phase data of instantaneous wavefront.Moreover,the data were introduced into Zernike polynomials in polar coordinates to obtain optical transfer functions and light field distribution of focal plane and defocus plane.Furthermore,the regularization constraints were used to restore the objective function and the cost function of the difference method.Finally,the minimum value of likelihood function was solved.Thus,the remote sensing degraded image restoration was achieved.Experimental results show that the average gray gradient,Laplace gradient modulus and Fisher information of the im-age restored by the proposed method are all greater than the values of original image.In addition,the restored image has richer detail features,higher resolution and better restoration.