局部多项式回归的图像重建改进

针对随机取样产生的低分辨率图像的分布特性,在局部多项式回归的数学模型下,利用经典核回归方法重建图像的过程中,引入与取样窗范围内随机取样像素点数目相关的权重项。改进的权重函数同时受取样窗范围的控制,更具适应性,利用可控核回归方法引入局部方向信息提高重建图像的局部结构特征。仿真结果表明,改进后的方法不仅提高了重建图像的分辨率,而且降低了重建图像的均方误差。

Research on Improvement of Local Polynomial Regression Image Reconstruction

Considering the distribution of random sampled low resolution image,based on the mathematical model of local polynomial regression,the classical kernel regression(CKR)method is used for image reconstruction.The weight function is adapted in the process and adaptively confined to the number of the pixels in the scale of sample window.Then steering kernel regression(SKR)method with local orientation is used to adapt better to local characteristics of the reconstruction image.Simulation results demonstrate the...