针对点扩散函数估计(PointSpreadFunction,PSF)不准确导致在气动退化图像盲复原处理后易造成振铃效应的问题,提出了一种改进的PSF估计方法,并将之应用于气动退化图像的复原处理中。具体做法是:构建平移不变的非下采样小波变换(NSWT),从而使之更易于处理图像奇异信息;基于小波变换模极大值和点扩散函数方差之间的关系,提出基于冗余提升NSWT的PSF估计方法;最后,将之应用于气动退化图像的盲复原中进行仿真验证。实验结果表明,使用改进后的PSF估计方法复原后的图像振铃效应等伪像程度显著减轻,图像质量明显改善,从而验证了构建的平移不变NSWT及改进PSF估计方法的有效性。
To solve the problem of severe tinging effect in aero-degraded image owing to imprecise estimation of Point Spread Function (PSF), an improved estimating function of PSF was put forward. First, translation variance Nonsubsampled Wavelet Transform (NSWT) was constructed for easily disposing singular information. Then, a method for estimation of PSF based on translation variance NSWT was bring forward by using the relation between maximum modulus of the wavelet transform and variance of PSF. Finally, it was used in blind restoration experiments of aero-optics degraded image for validation. The experiment results show that the image restored by the improved PSF estimation method has obviously weak ringing effect, and the quality of image is improved greatly, which proves the effectiveness of translation variance NSWT and the improved PSF.