基于影像序列的非均匀采样多尺度小波超分辨率影像重建

Multi-scale Wavelet Super-resolution Image Reconstruction Based on Ununiform Sampling Image Frames

原始影像序列间的非均匀采样以及影像的先验知识不足等是影响影像重建的难点问题。本文在小波理论基础上,阐述了空间域不同尺度的小波分解与影像高分辨率重建的关系,并根据非均匀采样数据对不同尺度空间的小波系数进行估算,然后利用估计的小波系数对高分辨率格网上的未知点进行插值重建,获取了高分辨率的影像。实验证明,本文提出的算法克服了先验知识不足和原始影像序列的非均匀采样的难题,并在获取的重建影像分辨率和清晰度方面有了较大提高,计算的复杂度改善也比较明显。

No-uniform sampling and short of knowledge of LR image sets are the main problems of image reconstruction technology. With the theory of wavelet, the paper describes the relationship between different spatial scale wavelet resolvabilities and FIR reconstruction. According to the given sample data, different scale wavelet moduli are estimated. Using the moduli, we reconstruct the unknown sapmple-points in FIR grids and get the FIR image at last. By experiment, it proves that the algorithm of the paper avoid the problems of image reconstruction mentioned above and greatly improves the image＇ s spatial resolution and definition. At the same time, the calculation complication is also reduced by using the reconstruction method.