【摘要】 提出融合方向扩散和高斯热扩散以及灰度保真项的偏微分方程图像放大模型,给出了数字实现方案,并应用到毫米波图像放大中。通过对Lena图像实验表明,高斯热扩散可减少线性插值算法带来的锯齿现象,方向扩散保护图像边缘的连续性,灰度保真项可平滑方块效应及噪声,与线性插值和文献〔7〕相比,本文方法在均方误差、信噪比和边缘保护指数性能最好。对手枪毫米波图像放大2倍时,本文算法在手枪背景区域平滑噪声、抑制虚假边缘能力明显好于文献〔7〕算法。更多还原

【Abstract】 As Gaussian heat diffusion has ability in filtrating saw tooth phenomenon caused by linear interpolation,and directional diffusion does in keeping on edge,and gray fidelity restraint item can smooth blocking artifacts,and remove noise satisfactorily,a new image magnification method,which is based on partial differential equation(PDE) involving above three items,is presented in this paper.Then we propose a digital scheme for this algorithm and give an application in the image magnification of millimeter wave.Compared to the linear interpolation and the method in(7),the Lena image experimental results indicate that our method has a better performance than others in mean square error(MSE) and signal to noise(SNR) as well as edge keeping index(EKI) used as some quantitative measures of images’ quality.Besides,the experiment on the handgun millimeter wave image which was magnified 2 times shows that our method has also a better ability to remove noise and eliminate the false edge in background than the method in(7).更多还原