计算机工程与应用
計算機工程與應用
계산궤공정여응용
COMPUTER ENGINEERING AND APPLICATIONS
2014年
1期
156-158,210
,共4页
盲反卷积%非局部%总体变分%正则化
盲反捲積%非跼部%總體變分%正則化
맹반권적%비국부%총체변분%정칙화
blind deconvolution%on-local%total variation%regularization
针对没有完整先验知识的降质图像的多帧图像恢复问题,利用非局部总体变分正则化方法,提出了多帧图像盲反卷积问题求解的有效数值算法。该算法既能保持重建图像的边缘与纹理结构又能抑制相关噪声,而且能同时重建原始图像和相关的点扩展函数PSF。实验结果表明提出的方法具有明显的优越性。
針對沒有完整先驗知識的降質圖像的多幀圖像恢複問題,利用非跼部總體變分正則化方法,提齣瞭多幀圖像盲反捲積問題求解的有效數值算法。該算法既能保持重建圖像的邊緣與紋理結構又能抑製相關譟聲,而且能同時重建原始圖像和相關的點擴展函數PSF。實驗結果錶明提齣的方法具有明顯的優越性。
침대몰유완정선험지식적강질도상적다정도상회복문제,이용비국부총체변분정칙화방법,제출료다정도상맹반권적문제구해적유효수치산법。해산법기능보지중건도상적변연여문리결구우능억제상관조성,이차능동시중건원시도상화상관적점확전함수PSF。실험결과표명제출적방법구유명현적우월성。
In order to solve multi-frame images restoration problem without complete prior knowledge, a non total varia-tion regularization method is applied to reconstruct original image and the associated Point Spread Function(PSF), a effi-cient numerical scheme is proposed to solve the variational problem. The proposed numerical scheme can preserve sharp edges and fine structures, textures of images while alleviating the concerns of noise amplification. Numerical experimen-tal results show the excellent performance of the proposed method.