基于α-expansion的超分辨率图像重建  

Image super-resolution reconstruction based onα-expansion algorithm

作　　者：张东晓[1,2] 李翠华[1] 

机构地区：[1]厦门大学信息科学与技术学院,福建厦门361005 [2]集美大学理学院,福建厦门361021

出　　处：《中国科技论文》2015年第8期937-941,共5页China Sciencepaper

基　　金：国家自然科学基金资助项目(61373077);国防基础科研计划(B0110155);国防科技重点实验室基金资助项目(9140C30211ZS8);高等学校博士学科点专项科研基金资助项目(20110121110020);福建省重点项目(2014H0034);黄慧贞集美大学学科建设基金资助项目(ZC2014010)

摘　　要：为了获得更精确的超分辨率重建结果,在重建高分辨率像素时,剔除对重建贡献微弱甚至没有贡献的低分辨率像素,保留有所贡献的低分辨率像素。对低分辨率像素的贡献进行程度化,获得所有低分辨率像素的加权贡献,并进一步建立超分辨率重建的极小化能量函数,使用α-expansion算法求解能量函数极小化问题。在求解过程中,保留3个像素间的关系,可避免由能量函数的近似而引入的错误信息。实验结果表明,所提方法是合理、有效的。In order to obtain more accurate super-resolution reconstruction results, the low-resolution pixels are deleted, which present few or none contribution to the reconstruction of a high-resolution pixel. Grading method is introduced to weight the con- tribution of these remainder low-resolution pixels. Minimization energy function for super-resolution is constructed. In addition, -expansion is chosen to minimize the energy function. Triple priors (the relation between three adjacent pixels) are retained in the solving process, which restrain error messages introduced by energy function over approximation. The experiment results show that the proposed method is both reasonable and effective.

关 键 词：图像处理 超分辨率 α-expansion算法 图像重建 程度化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]