基于感知去模糊的高分辨率破损图像修复方法
High Resolution Damaged Inpainting Method Based on Perceptual Deblurring
濮毅1
作者信息
- 1. 马鞍山职业技术学院电子信息系,安徽 马鞍山 243031
- 折叠
摘要
为提高修复图像的视觉效果,提出基于感知去模糊的高分辨率破损图像修复方法.将高分辨率图像样本输入卷积自编码生成对抗网络中,利用编码器降维处理并输出其低维特征矩阵后,由解码器对其升维并解码,最终采用生成器完成映射学习.通过不断搜寻,获得与输入高分辨率图像L1 距离差异最小的生成图像,由判别网络对其作真假判断,实现高分辨率破损图像的粗修复后,再将其输入感知去模糊网络模型中,增强图像细节信息后实现高分辨率图像修复.实验结果表明:该方法修复后的高分辨率图像细节丰富、颜色自然、视觉效果突出.
Abstract
To improve the visual effect of repaired images,a high-resolution damaged image restoration method based on perceptual deblurring is proposed.Using high-resolution image samples as the input for convolutional self encoding to generate adversarial networks,the encoder is used to reduce the dimensionali-ty and output its low dimensional feature matrix.After that,the decoder increases the dimensionality and de-codes it.Finally,the generator is used to complete mapping learning.By continuously searching for the generated image with the smallest distance difference from the input high-resolution image L1,the discrimi-nant network determines whether it is true or false.After the coarse repair of the high-resolution damaged image,it is used as input to the perceptual deblurring network model to enhance image detail information and achieve high-resolution image restoration.The experimental results show that the high-resolution images repaired by this method have rich details,natural colors,and outstanding visual effects.
关键词
感知去模糊/高分辨率破损图像/特征矩阵空间/粗修复/L1距离差异/真假判断Key words
perceptual deblurring/high resolution damaged image/feature matrix space/rough re-pair/L1 distance difference/true or false judgment引用本文复制引用
基金项目
安徽省教育厅高等学校省级质量工程项目(2022jyxm1577)
安徽省教育厅高等学校省级质量工程项目(2022cjrh046)
安徽省教育厅高等学校省级质量工程项目(2021jxtd286)
出版年
2024
NETL
NSTL
万方数据