基于加权低分辨率图像的自适应正则化图像重建  

Adaptive Regularization Image Reconstruction With Weighted Low-resolution Images

作　　者：王抒[1] 齐苏敏[1] 

机构地区：[1]曲阜师范大学计算机学院,山东曲阜273165

出　　处：《软件导刊》2012年第4期161-163,共3页Software Guide

基　　金：山东省软科学项目(2011RKGA2020)

摘　　要：图像重建是一个病态问题,需要应用逆过程获得原始图像的近似估计。亚像素配准虽在图像重建过程中发挥了重要作用,但难以获得准确值。提出一种自适应图像重建迭代算法和基于加权低分辨率图像的Tikhonov正则化参数的自适应估计方法。权重系数保持了每个低分辨率图像的逼真度,而正则化系数则控制了图像平滑度。实验结果表明,此算法无论在客观测量还是在视觉评价上,都优于传统的Tikhonov正则化方法。The image reconstruction is a ill-posed problem and requires to apply an inverse procedure to obtain an approximate estimation of the original image. In practice, Subpixel registration which plays an important role in the image reconstruction process is difficult to get accurate value. In this paper, we propose an adaptively iterative image reconstruction algorithm and simultaneous adaptive estimation of the Tikhonov regularization parameter based on the weighted low resolution images according to the multiple parameter generalization of Tikhonov-Miller regularization. The weight coefficients act as the cross-channel fidelity to each low-resolution image, while the regularization parameter controls smoothness term using Tikhonov regularization. Experimental results indicate that the proposed algorithm outperforms conventional Tikhonov regutarization approaches in terms of both objective measurements and visual evaluation.

关 键 词：图像重建 亚像素配准 正则化方法 高斯噪声 迭代 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]