摘要
基于模糊退化过程的空不变卷积模型,依据跳跃边缘的不连续性在总变分最小化过程中的指导性作用,提出一种在已知模糊核情况下,基于Richardson-Lucy和总变分(Total Variation)正则化的两步非盲模糊图像复原方法。Richardson-Lucy算法得到清晰边缘的初始估计,然后以边缘的不连续性指导总变分能量泛函的优化,求得具有清晰突出边缘的复原图像。本算法既弥补了Richardson-Lucy算法由于缺乏正则约束而带来了收敛性和噪声放大等方面的不足,同时对总变分正则项采用了上界凸函数最小化的优化方法,避免了传统总变分最小化时非线性偏微分方程的求解问题,降低了算法复杂性。经过真实模糊图像测试表明,算法能保持恢复图像的整体视觉效果及小尺度的边缘结构,并且复原效果好于基于总变分的Richardson-Lucy迭代复原算法。
Based on empty constant convolution model in fuzzy degradation process,a new deconvolution method combing Richardson-Lucy and Total variation regularization to restore a sharp image is proposed,according to the guidance of jump edge discontinuity in total variation minimization process.The Richardson-Lucy is used to get the initial estimate of clear edge,then edge discontinuities are used to guide the total variational energy functional optimization to get restoration image with outstanding and clear edge.This method makes up the shortfall of Richardson-Lucy＇s convergence and noise amplification which caused by lack of regular constraintion,and it also avoid the problem of solving nonlinear partial differential equations in traditional total variation minimization,using the method of upper convex minimization optimization,so this method can reduce the complexity of the algorithm.The test shows that this method can maintain and recovery the overall visual effect and small-scale edge structure of the image,and its restoration results are better than Richardson-Lucy method.
出处
《电子测量技术》
2011年第12期45-48,共4页
Electronic Measurement Technology
