摘要
在高光谱图像混合噪声恢复过程中,传统全变分方法仅考虑了图像的局部平滑性,而未能考虑高光谱图像的非局部相似性,因此恢复的图像的质量有待进一步提升。针对这一问题,考虑到高光谱图像的非局部相似性,引入非局部全变分。实验结果表明,通过将非局部全变分与张量低秩分解相结合的方法得到的结果较传统变分方法在视觉效果和客观评价指标方面都有较大提升。
In the process of hyperspectral image restoration with mixed noise,the traditional total variation method only considers the local smoothness of the image,but fails to consider the nonlocal similarity of the hyperspectral image,so the quality of the restored image needs to be further improved.To solve this problem,considering the nonlocal similarity of hyperspectral images,nonlocal total variation is introduced.The experimental results show that the results obtained by combining nonlocal total variation with tensor low rank decomposition are better than those obtained by traditional variation methods in terms of visual effect and objective evaluation index.
作者
孔祥阳
王惠
李欣星
伍晓亮
Kong Xiangyang;Wang Hui;Li Xinxing;Wu Xiaoliang(Sichuan Engineering Technical College,Deyang,Sichuan,618000,China)
出处
《装备制造与教育》
2021年第1期44-46,共3页
Equipment Manufacturing and Education
基金
四川工程职业技术学院2020年度院级科研项目:基于张量分解的遥感图像恢复技术研究(编号:YJ2020KJ-14)阶段性研究成果。
关键词
非局部全变分
张量
低秩分解
混合噪声
nonlocal total variation
tensor
low rank decomposition
mixed noise
