基于非局部先验红外运动模糊图像复原方法
A non-local prior of infrared motion blurred image restoration method
何易德 1朱斌 1汤磊 1蒲小平 2王升哲 1代辉 1郭志伟 1王捷1
作者信息
- 1. 西南技术物理研究所,成都 610046,中国
- 2. 中国人民解放军93128部队,北京 100843,中国
- 折叠
摘要
为了实现红外运动模糊图像复原,采用了基于红外图像的非局部稀疏先验约束建模方法.通过分析红外运动模糊成像特征,在盲反卷积框架下,提出了一种基于运动信息的图像非局部稀疏先验约束建模方法,通过计算图像的运动模糊核,进而复原运动模糊图像.结果表明,所提出的基于运动信息的图像非局部稀疏先验约束方法,针对性强,能有效地复原运动幅值较大的红外运动模糊图像;概率模糊检测、结构相似度和峰值信噪比均有不同程度的提高,尤其是峰值信噪比提高接近8％,且运动幅值越大,复原结果越明显.本研究为红外成像系统的应用打下了基础.
Abstract
In order to restore motion degradation blur of the strap-down guidance infrared seeker,a non-local sparse prior constraint modeling method for infrared images was proposed.By analyzing the infrared motion blur imaging features of the strap-down platform,a non-local sparse prior constraint modeling method based on motion information was proposed in the blind deconvolution framework,which can estimate the motion blur kernel of the image and restore the infrared motion blur image.The result shows that the non-local sparse prior constraint method based on motion information proposed in this paper is highly targeted and can effectively restore infrared motion blurred images with large motion amplitudes.Cumulative probability of blur detection,structural similarity,peak signal-to-noise ratio all show varying degrees of improvement,especially peak signal-to-noise ratio increases by nearly 8％,and the larger the motion amplitude,the more obvious the restoration results.This study lays the foundation for the application of the strap-down guidance infrared imaging system.
关键词
图像处理/红外运动模糊图像复原/运动成像特征/非局部稀疏先验Key words
image processing/infrared motion blurred image restoration/motion imaging features/non-local sparse prior引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据