目标边缘保持约束的图像联合解卷积复原算法

Joint deconvolution image restoration algorithm based on the edge-preserving model

针对图像复原过程中去噪与目标边缘特征保持之间的矛盾,该文提出了一种基于目标边缘保持的图像联合解卷积复原算法。首先,构建了一个目标边缘保持约束模型,实现对图像小梯度特征(噪声为主)的平滑、对图像大梯度特征(目标边缘为主)的保留,平衡复原处理过程中图像去噪与目标边缘保持之间的矛盾;然后,将目标边缘保持约束先验模型,引入MAP图像复原框架,提升MAP复原算法的可靠性和收敛性;最后,利用共轭梯度迭代优化计算过程,加快算法收敛速度。实验结果表明,该算法能较好地平衡图像去噪与目标边缘特征保持之间的矛盾,实现了图像高清晰复原。

In view of the contradiction between image denoising and edge-preserving,a new image restoration algorithm based on the edge-preserving model was proposed.First,a target edge-preserving constraint model was constructed for smoothing image small gradient features（mainly noise）,retaining image large gradient features（edges）,and balancing the contradiction between image denoising and edgepreserving.Then,the target edge-preserving model was introduced into the MAP image restoration framework to improve the reliability and astringency of the MAP algorithm.Finally,the conjugate gradient iteration was used to optimize the calculation,and the convergence speed of the algorithm was accelerated.The experiment results of image restoration proved that the new algorithm could effectively balance the contradiction between image denoising and edge-preserving,and it did well in image restoration.