摘要
讨论了从一组低采样降质的视频图像重建超分辨率图像中未知参数的估计问题.使用L-Curve标准来估计正则化参数.然而,L-Curve的计算代价十分昂贵.它需要计算正则化近似解和残差的范式.为此提出一种基于Lanczos算法和Gauss积分理论的算法,在超分辨率图像重建中的参数估计中可以减少L-Curve的计算代价.
this paper considers the estimation of the unknown parameter for the problem of reconstructing a super-resolution image from multiple under-sampled, shifted, and degraded frames. L-curve criterion is used to estimate the regularization param eters. However, the computation of the L-curve is quite costly for large problems. The determination of a point in the L-curve requires that both the norm of the regularized approximate solution and the norm of the correspondlngresidual vector are available. The paper proposes an approximate technique based on the Lanczos algorithm and Gauss quadrature theory, which can reduce computational complexity of the L-curve for the efficient parameter estimation in high-resolution image reconstruction.
出处
《小型微型计算机系统》
CSCD
北大核心
2006年第6期1120-1123,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60075010)资助
黑龙江省教育厅科学基金项目(10551115)资助
哈尔滨师范大学科研基金项目(KM2005-11)资助