节点文献
无纹理图像修复模型的快速数值算法
Fast Algorithm for Non-texture Image Inpainting Model
【摘要】 将曲率驱动扩散(CDD)模型应用于无纹理图像的修复,对其数值离散格式进行改进,计算分为两步执行:在修复区域内采用时间步进法;修复区域外利用加性算子分裂方法,该方法加快了收敛速度,数值实验结果表明效果较好.
【Abstract】 The curvature-driven diffusions(CDD) model was applied to the non-texture image inpainting.The numerical discretization scheme was developed.The computation is constituted of two steps.For inside the inpainting domain,we employed the time match method;while for outside the inpainting domain,we employed the addtive operator splitting(AOS) method,which can gain an accelerated convergence.The experiment results are satisfying.
【关键词】 图像修复;
无纹理图像;
加性算子分裂方法;
【Key words】 image inpainting; non-texture image; additive operator splitting(AOS) method;
【Key words】 image inpainting; non-texture image; additive operator splitting(AOS) method;
【基金】 国家自然科学基金(批准号:J0730101)
- 【文献出处】 吉林大学学报(理学版) ,Journal of Jilin University(Science Edition) , 编辑部邮箱 ,2011年02期
- 【分类号】TP391.41
- 【被引频次】1
- 【下载频次】140