在小波变换域内实现图像的超分辨率复原

提出了在小波域内实现图像的超分辨率复原的方法，这种方法可以达到自适应边缘保持的目的，算法特点如下：(1)对观测模型实施正交小波变换，获得超分辨率复原问题的空频域描述；(2)采用广义高斯概率模型来构建超分辨率图像的尺度系数和小波系数的先验描述；(3)采用半二次正则化迭代方法来完成小波域超分辨率复原的求解过程。

Super-Resolution Restoration from Image Sequences in the Wavelet Domain

In this paper, a wavelet domain super resolution restoration method has been proposed, which can be used to estimate the wavelet coefficients of the high resolution image. An adaptive edge preserving image prior is also assumed for the data in this method. The proposed algorithm is characterized by follows: First, the spatial frequency representation of the super resolution restoration problem is obtained by processing the observation model with the orthogonal wavelet transform. Then, the prior model for the wavelet coefficients of the high resolution image is assumed to be i.i.d generalized Gaussian distribution. Finally, the wavelet domain super resolution restoration is accomplished through half quadratic regularization iteration algorithm.