一种离焦模糊图像的约束复原方法

An algorithm of constrained restoration for defocus blurred image

针对离焦模糊图像,提出了一种基于图像质量的约束复原算法。该算法基于平滑约束最小二乘,采用正则化技术以减少高频噪声的放大。使用人类视觉系统(HVS)特性中的对比敏感度函数(CSF)对图像功率谱加权来评价图像质量,通过搜索图像质量的最大值实现对正则参数的自适应选择,使复原结果更符合人的视觉感受。实验结果表明:该算法不需要估计降质图像的噪声水平,却能获得和平滑最小二乘复原方法相当甚至更好的复原结果。

As for defocus blurred image, a constrained restoration algorithm was proposed based on image quality. This algorithm used regularization to eliminate the noise ampliation. The contrast sen- sitivity function （CSF） of the characteristics of human visual system （HVS） was used for weighting the image power spectra to evaluate the image quality. The regularization parameter was auto selected by searching the maximum of the image quality in order to make the resorted result fit human perception more suitably. The experiments show that the proposed algorithm does not need to estimate noise level and can achieve good or better result as compared with the smoothing constrained least square restoration.