摘要
针对传统算法在图像边缘细节提取不够准确和估测的模糊核与真实模糊核有很大差距的问题,提出一种局部块相对梯度的图像复原算法.通过双边滤波器和冲击滤波器进行降噪和边缘增强后用相对梯度的方法将模糊图像进行边缘提取,基于自然图像的梯度分布约束先验知识,结合疏导滤波的思想,在取得模糊核后对复原的图像进行相关约束.利用双三次内插值的方法将复原的图像进行放大,恢复到原模糊图像的尺寸,并与其他复原算法进行客观评价对比实验.结果表明,基于疏导滤波和梯度先验分布算法对图像盲复原效果有一定提高作用.
A local block relative gradient image restoration algorithm is proposed to address the issues of inaccurate edge detail extraction and significant gap between estimated fuzzy kernels and true fuzzy kernels in traditional algorithms.After denoising and edge enhancement using bilateral filters and impulse filters,the blurred image is edge extracted using relative gradient method.The gradient distribution constraint prior knowledge based on natural images is introduced,combined with the idea of grooming filtering.After obtaining the fuzzy kernel,relevant constraints are applied to the restored image.Using the bicubic interpolation method to enlarge the restored image and restore it to the size of the original blurred image,and conducting objective evaluation and comparison experiments with other restoration algorithms.The results indicate that the algorithm based on grooming filtering and gradient prior distribution has a certain improvement effect on blind image restoration.
作者
郭锋锋
GUO Fengfeng(Department of Computer and Information,Suzhou Vocational and Technical College,Suzhou 234000,Anhui,China)
出处
《山西师范大学学报(自然科学版)》
2025年第1期51-57,共7页
Journal of Shanxi Normal University(Natural Science Edition)
基金
安徽省教育厅自然科学重点研究项目(2022AH052768)
安徽省教育厅自然科学重点研究项目(2023AH052958)
安徽省教育厅计算机应用技术专业教学创新团队(2022cxtd161)。
关键词
图像盲复原
相对梯度
模糊核估计
疏导滤波
blind image restoration
relative gradient
fuzzy kernel estimation
grooming filtering
