| 基于最小鉴别信息的正则化图像恢复方法 |
| |
| 引用本文： | 吴显金,王润生.基于最小鉴别信息的正则化图像恢复方法[J].计算机工程与设计,2007,28(3):605-607. |
| |
| 作者姓名： | 吴显金 王润生 |
| |
| 作者单位： | 国防科技大学ATR国家重点实验室,湖南,长沙,410073 |
| |
| 摘 要： | 提出在正则化图像恢复方法中将图像恢复结果与先验图像的最小鉴别信息作为新的正则化约束.同传统的正则化约束不同,新的约束使得恢复的图像与给定的先验图像具有最相似的灰度分布.同时给出一种自适应确定正则化参数的方法.实验结果表明,新方法在恢复效果上要优于传统的正则化方法,但对噪声则比较敏感.因此,提出在降质图像含有较多的噪声时保留传统的正则化约束,以达到更好的恢复效果.
|
| 关 键 词： | 图像恢复 正则化约束 先验图像 最小鉴别信息 正则化参数 最小 鉴别信息 正则化方法 图像恢复 恢复方法 discrimination information minimum based 敏感 比较 噪声 效果 实验 正则化参数 自适应 灰度分布 相似 约束 结果 |
| 文章编号： | 1000-7024（2007）03-0605-03 |
| 修稿时间： | 2006-01-23 |
Regularized image restoration based on minimum information discrimination |
| |
| WU Xian-jin,WANG Run-sheng.Regularized image restoration based on minimum information discrimination[J].Computer Engineering and Design,2007,28(3):605-607. |
| |
| Authors: | WU Xian-jin WANG Run-sheng |
| |
| Affiliation: | ATR National Laboratory, National University of Defense Technology, Changsha 410073, China |
| |
| Abstract: | The minimum information discrimination between restored image and prior image is proposed as new regularization constraint in the regularized image restoration.Different from the classical regularization constraint,the new constraint leads to the restored image is similar to the prior image in gray-value distribution.At the same time,the method to choose regularization parameter adaptively is given.The experimental results show the new approach is better than the classical regularization approach in restored effect but it is sen-sitive tothe noise.Therefore,the classicalregularization constraint is preserved to get better restored effect in the presence ofmuch noise. |
| |
| Keywords: | image restoration regularization constraint prior image minimum information discrimination regularization parameter |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
