摘要
通过各向同性的格子波尔兹曼模型模拟图像修复的过程,通过将图像的曲率和梯度等几何信息嵌入到LBM微观演化方程的松弛因子中,实现曲率驱动的各向同性扩散以达到图像修复的目的。实验结果表明,LBM图像修复方法可以得到很好的修复结果,且修复效率优于经典的BSCB模型和TV模型算法。
We propose a new isotropic diffusion LBM model to embed the geometric features of the image into the relaxation parameters of the Lattice Boltzmann Equation for non-texture inpainting.Experiment results show that the LBM based inpainting method has high efficiency than the well known methods such as BSCB and TV while producing comparable results in terms of visual inspection.
出处
《电子测量技术》
2011年第3期46-48,65,共4页
Electronic Measurement Technology
关键词
图像修复
格子波尔兹曼方法
各向同性扩散
non-texture inpainting
lattice boltzmann method
isotropic diffusion
