多方向加权TV和变换域非局部正则化图像去模糊

发布时间：2018-01-30 12:32

本文关键词： 图像去模糊多方向加权TV 变换域非局部正则化交替方向法　出处：《天津大学学报(自然科学与工程技术版)》2017年08期 　论文类型：期刊论文

【摘要】：针对传统全变差(TV)去模糊对噪声敏感且细节恢复能力有限等缺点,利用边缘检测对传统TV模型进行改进,并受空域非局部自相似性正则化思想启发,将图像的变换域非局部自相似性约束融入去模糊模型,提出一种基于边缘检测的多方向加权TV和变换域非局部正则化的图像去模糊方法.首先,运用边缘检测将中心像素邻域内的像素对划分为同侧像素对和异侧像素对,对不同类型的像素对采用不同的权重,在去模糊的同时尽可能保持图像边缘等细节特征;其次,为充分利用先验信息,将变换域非局部正则化约束融入到改进的TV模型,进一步改善图像视觉质量;最后,对新模型进行有效求解.实验结果表明,本文算法在去模糊的同时可更好地保留图像的边缘、纹理等细节特征.
[Abstract]:Aiming at the shortcomings of traditional total variation (TVV) deblurring, which is sensitive to noise and limited in detail recovery ability, the traditional TV model is improved by edge detection, which is inspired by the idea of spatial non-local self-similarity regularization. An image de-blurring method based on multi-direction weighted TV and non-local regularization based on edge detection is proposed by incorporating the non-local self-similarity constraints in the transform domain into the de-blurring model. Using edge detection, the pixel pairs in the center pixel neighborhood are divided into the same side pixel pair and the different side pixel pair, and the different types of pixel pairs are given different weights. While deblurring, the image edge and other details are kept as much as possible; Secondly, in order to make full use of the prior information, the non-regularization constraints in the transform domain are incorporated into the improved TV model to further improve the visual quality of the image. Finally, the new model is solved effectively. The experimental results show that the proposed algorithm can better preserve the edge and texture features of the image while removing the blur.
【作者单位】：天津大学电气自动化与信息工程学院;
【基金】：国家自然科学基金资助项目(61472274)~~
【分类号】：TP391.41
【正文快照】： 图像去模糊是计算机视觉、图像处理领域的研究热点,因其具有前沿性、应用广等特点而备受关注.在众多去模糊方法中,全变差(total vaviation,TV)正则化因其较好的边缘保持能力被广泛用于图像去噪、图像去模糊等领域[1-2],但是其细节和纹理恢复能力有限.因此,出现了很多改进方法,

【相似文献】