[1]袁小华,等.改进的有参超分辨率图像盲恢复[J].南京理工大学学报(自然科学版),2006,(03):343-347.
 YUAN X iao-hua,GAO X iu-m ei,et al.Improved Parametric Blind Super-resolution Image Restoration[J].Journal of Nanjing University of Science and Technology,2006,(03):343-347.
点击复制

改进的有参超分辨率图像盲恢复
分享到:

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

卷:
期数:
2006年03期
页码:
343-347
栏目:
出版日期:
2006-06-30

文章信息/Info

Title:
Improved Parametric Blind Super-resolution Image Restoration
作者:
袁小华1 2 高秀梅3 夏其英4 夏德深1
1. 南京理工大学计算机科学与技术学院, 江苏南京210094; 2. 上海农科院农业信息化工程技术中心, 上海201106; 3. 淮阴师范学院计算机科学系, 江苏淮安223001; 4. 临沂师范学院化学系, 山东临沂276005
Author(s):
YUAN X iao-hua1 2 GAO X iu-m ei3 XIA Q -i y ing4 X IA De-shen1
1. Schoo l of Computer Science and Techno logy, NUST, N anjing 210094, China; 2. Ag ricultural Informat ion Eng ineering and Techno logy C enter, ShanghaiA cademy of Agricu ltural Sc iences, Shanghai 201106, Ch ina; 3. Department o fC ompu ter Science, Huaiyin Teachers Co llege, Hua ian 223001, China; 4. Department of Chem istry, L iny iNormalUn iversity, L iny i 276005, China
关键词:
超分辨率图像盲恢复 交替优化 Neumann边界 预处理器 边界修正
Keywords:
b lind super-reso lution image resto ration a lternative optim ization Neum ann boundary preconditioner boundary modification
分类号:
TP 391. 41
摘要:
改进了有参超分辨率图像盲恢复中的边界与正则处理以及模糊参数估计方法,用交替迭代优化来估计模糊参数和高分辨率图像,用Neumann边界图像模型和修改边界像素来避免边界错误,用近似重排系统矩阵的带结构矩阵,来构造求解高分辨率图像的预处理共轭梯度算法。人工退化图像序列上的实验结果表明了改进方法的有效性。
Abstract:
To improve the boundary condition processing, the regu larizat ion processing, and the b lurring parameter est imation of parametric blind super-reso lution image resto ration, an a lternative iteration optim ization is used to estimate the im age b lurring and high-reso lution image. The neumann boundary cond ition and the boundary p ixe lmodifica tion of the low er-reso lu tion image are adopted to avo id the boundary errors. In so lving the high-reso lut ion image, them atrixes close to the reordered system matrix and w ith spec ia l structures are used to make the precond it ioner in the precond it ioned conjugate gradient a lgorithm. The test results using the synthetic low-reso lut ion sequences are presented to show the validation o f the proposed improvements.

参考文献/References:

[ 1] Nguyen N, M ilanfa P, Go lub G. E fficent genera lized cross-validation w ith applica tions to pa rame tric im ag e restoration and reso lution enhancem en t [ J ]. IEEE Transactions on Image Process ing, 2001, 10 ( 9): 1 299- 1 308.
[ 2] NgK M, Chan R H, Chan T F, e t a .l Cosine transfo rm precond itioners fo r high resolution im age reconstruction [ J]. L inear A lgebra and Its Applica tions, 2000, 316 ( 1): 89- 104.
[ 3] Buck ley M J. Fast com putation o f a discretized th inplate sm oothing sp line for im age data [ J]. B iom etrike, 1994, 81 ( 2) : 247- 258.
[ 4] E ld?n L. A w eighted pseudo inverse, genera lized singula r va lues, and constra ined least squares prob lem s [ J]. B IT, 1982, 22: 487- 502.
[ 5] Jaehoon K, Bose N K. Spatia l resto ra tion w ith reduced bounda ry erro r [ A ]. In Proc M athem atical Theory o f Netwo rks Band System s (MTNS ) [ C ] . South Bend: University of Notre Dame, 2002. 1- 6.
[ 6] S ignal & image processing institute, E lectr ica l Eng-i neer ing Departm ent, Un ive rsity of southern Ca lifo rn ia. H ouse im age. [ EB /OL ]. http: / / sip.i usc. edu /services /da tabase / database. cg?i vo lum e= m isc& image= 5 # top, 2004- 10- 28.

备注/Memo

备注/Memo:
基金项目: 香港特区政府研究资助局资助项目( CUHK /4180 /01E), 江苏省教育厅自然科学基金( 04KJD520037) 作者简介: 袁小华( 1969- ), 女, 四川三台人, 博士生, 主要研究方向: 数字图像处理, E-mail:yxiaoh@citiz. net; 通 讯作者: 夏德深( 1941- ), 男, 博士生导师, 博士, 主要研究方向: 遥感和医学图像分析与模式识别, Email:deshen_x@ 263. net。
更新日期/Last Update: 2006-06-30