基于统计特征的彩色图像快速插值方法

本文首先阐述了基于统计特征的图像插值方法,该方法通过提取待插入像素所在区域的协方差矩阵和协方差向量,得出适应于边缘位置和方向的插值权重.为了把基于统计特征的图像插值方法应用于彩色图像插值领域,本文提出了以下措施以提高计算速度:仅对Y图像估计插值权重,并同时应用到R、G、B三个分量的插值;对边缘像素应用基于统计特征的图像插值方法,而对非边缘像素应用简单的双线性插值,即混合图像插值方法.这些措施提高了计算速度,并保证了图像质量.实验表明了该算法在计算速度和插值图像质量方面的有效性.

Statistical Feature Based Fast Color Image Interpolation

This paper discusses a kind of statistical feature based image interpolation method at first.Using the covariance matrix and covariance vector in low-resolution image,one can obtain the interpolation coefficients,which can adapt edge orientation.In order to employ statistical feature base interpolation in the area of color image interpolation,this paper present the following methods to increase the calculating speed:estimating the interpolation coefficients only based on Y image,and applying them to R,G,and B images interpolation simultaneously;For edge pixels,statistical feature based image interpolation method is exploited,while for non-edge pixels,we only use simple bilinear interpolation.This method can be called mixture interpolation.Benefit from these measures,one can get high calculating speed while keeping the image quality unchanged.The experiments show that this method is both effective for calculating speed and interpolated image quality.