摘要
针对TV模型修复算法只沿梯度垂直方向扩散,容易在平滑区域引入阶梯效应,迭代效率低,易产生假边缘的缺点,分析比较了TV图像修复模型的性能,提出了一种改进的图像修复算法。该算法同时结合了各向同性和各向异性扩散,利用区域频率差异实现了在不同的区域使用不同的迭代方程,有效避免了原始算法引入的阶梯效应,提高了迭代效率。实验结果表明,该算法与TV模型算法相比,在具有同样修复效果的前提下,避免了阶梯效应并优于TV模型的修复速度。
TV image inpainting only repairs algorithm along the vertical direction,making diffusion gradi- ent in smooth area into stair step effect, reducing iterative efficiency,producing false edge faults. We ana- lyse and compare the TV image restoration model and repair the performance of the model is put forward in this paper. Puts forward an improved image restoration algorithm,the algorithm in combination with the isotropic and anisotropic diffusion, the use of regional frequency difference in different area realized the use of different iterative equations, effectively avoid the original algorithm introduces ladder effect,to improve the iterative efficiency. The experimental results show that the algorithm and the TV algo- rithm, have the same effect on the premise of repair, the algorithm avoids the stair step effect at the same time with the model is better than TV repair speed.
出处
《吉林大学学报（信息科学版）》
CAS
2012年第1期72-77,共6页
Journal of Jilin University（Information Science Edition）
关键词
图像修复
各向同性
各向异性
TV模型
阶梯效应
image inpainting
isotropic diffusion
anisotropy diffusion
TV model
step effect
