【作者】 邹才凤； 钟似玢； 罗笑南； 潘琪；

【Author】 ZOU Cai-feng ZHONG Si-fen LUO Xiao-nan PAN Qi (School of Information Science and Technology,Sun Yat-sen University,Guangzhou,510275) (Key Laboratory of Digital Life(Sun Yat-sen University),Ministry of Education,Guangzhou,510275)

【机构】 中山大学信息科学与技术学院； 中山大学数字家庭教育部重点实验室；

【摘要】 为了获得高质量的插值图像,本文提出了基于 ENO(Essentially Non-Oscillatory,基本无振荡)思想的自适应的纯二维图像插值方法。求出每个模板对应的拟合曲面在待插值点处的梯度幅度,选择梯度幅度最小值对应的模板,即最光滑的模板,来计算待插值点处的灰度值。实验结果表明,本文方法插值得到的图像边缘比较清晰,同时相对于一维 ENO 的张量积插值方法,锯齿效应有所改善。更多还原

【Abstract】 In order to obtain interpolated images with superior quality,the paper proposes an adaptive pure two- dimensional image interpolation method based on Essentially Non-Oscillatory(ENO).The gradient magnitudes are calculated at the point to be interpolated on the fitting surfaces of every stencil.The smoothest stencil with minimal gradient magnitude is selected to calculate the gray scale value.Experimental results show that the method of the paper can produce interpolated images with sharp edges.In contrast to the tensor product method of one dimensional ENO,the new method can alleviate staircase effect.更多还原