基于交替方向乘子法的图像盲复原

为了克服正则化理论的全变分图像盲复原模型中出现的运行效率低、效果不好等问题,提出一种基于交替方向乘子法的盲复原迭代算法。该算法通过交替迭代的方式,将复原图像与点扩散函数交替估计,同时不必更新惩罚项从而提高了运行速度和复原的质量。计算同时加入了对点扩散函数的归一化和阈值约束条件以及对图像的正定性条件。数值试验中,对不同模糊类型的图像进行了盲复原处理,并与已有的其他盲复原方法进行了比较。从主观评价能够发现,提出的算法能够改进图像的质量,提高其分辨率;通过客观指标比较,峰值信噪比(peak signal to noise ratio, PSNR)最大能够提高1.2 dB,结构相似度(structural similarity index, SSIM)最大提高1%,计算时间最大节约一半左右。

Blind image restoration using alternating direction method of multipliers

In order to overcome the low operating efficiency and poor reconstruction quality in the total variation blind image restoration model of the regularization theory, an iterative algorithm of blind restoration based on alternating direction method of multipliers algorithm was proposed. The restored image and the point spread function were estimated alternatively by alternating iteration to improve the running speed and reconstruction quality through a way without updating the penalty term. The normalization and threshold constraint condition of the point spread function, and the positive definite condition of the image were added while calculating. In the numerical experimentation, the blind restoration of the images with different fuzzy types were carried out, and it was compared with other existing blind image restoration methods. The proposed algorithm could improve the quality and the resolution ratio of the image. Through objective comparison, the peak signal to noise ratio of the proposed algorithm could be increased by 1.2 dB at most,the average structural similarity was increased maximumly by 1% and the computation time was saved maximumly by about half.