摘要
根据矢量图像像素值特点,利用偏微分方程理论中的热传导数学模型,提出了一种基于非线性各向异性扩散PDE的矢量图像放大法。实验结果表明:文中所提出的算法对灰度图像和彩色图像进行放大均有较好的视觉效果。与传统插值放大算法相比,放大后图像减轻了模糊化,锯齿化以及振铃化效应;与已有的基于偏微分方程的图像放大算法相比,该算法能够较好地保持图像的边缘特征和细节信息。
A model based on a nonlinear anisotropic diffusion PDE for the interpolation of vector-valued images is proposed by applying heat conduction model. Experimental results show that the proposed method both for gray level and for color images have better visual effect. Compared to classic interpolation methods, the proposed method increases the image qualities and reduces the undesirable artifacts, such as blurring, aliasing, ringing. The proposed method can maintain the image edge features and details compared with other existing PDE-based interpolation methods.
出处
《光学仪器》
2011年第5期25-29,共5页
Optical Instruments
关键词
偏微分方程(PDE)
非线性各向异性扩散
图像插值
矢量图像
partial differential equations （PDE）,nonlinear anisotropic diffusion
imageinterpolation
vector-valued images
