基于退化模型和邻域嵌套的彩色图像超分辨率自适应重建

Degradation model and neighbor embedding based color image adaptive super-resolution reconstruction

为有效提升图像质量,提出一种基于图像退化模型和邻域嵌套的彩色图像超分辨率重建算法.通过退化模型在彩色空间上得出图像超分辨率重建训练集,并根据此训练集进行图像邻域分块.为了在训练过程中抑制噪声并锐化图像中的边缘信息,提取训练集亮度和梯度特征并进行特征融合.为了有效提升重建算法的自适应性,引入图像重建优化参数和边缘信息参数,根据此参数给出重建模型.基于最小二乘法的基本思想得出重建图像分块,并根据对应的次序对分块进行组合,得出最终高分辨率图像.实验结果表明,该算法效果良好.

A new color image super-resolution reconstruction method base on degradation model and neighbor embedding is proposed to effectively improve the image quality.Through image degradation model training set is obtained in color space and then according to this set the image area is cut into patches.A new combination of features is proposed based on luminance and gradient,which can preserve edges and smoothen color regions.Then image super-resolution optimization parameter（SOP） and edge information parameter（EIP） are introduced in order to effectively enhance the adaption of the algorithm and according to which,SR（super-resolution） model is given.Finally,the specific process of patches by Least Squares method is obtained.And the patches are stitched according to the corresponding coordinates,where the overlapping regions of patches are averaged.Experiments show that the proposed algorithm performs better in both quantitative and qualitative evaluation.