摘要
为了提升数字图像修复算法的效果,改进了NSCT图像修复算法,提出领域方差法及领域能量法,更好地实现了待修复区域周围的像素信息提取,改进了优先权值公式,优化了图像修复顺序.通过实验结果及分析表明,与其他方法相比,改进算法取得了更好地修复效果,有效的克服了Criminisi算法中的锯齿效应.
In order to improve the effect of digital image inpainting algorithm, In this pa- per, NSCT image inpaint ing algorthm is improved, propose neighbourhood variance method and neighbourhood energy method, better implementation of the pixel information extrac- tion around the area to be repaired, improved formula of priority weight, optimized image inpainting sequence. The experimental results and analysis show that, compared with other methods, the improved algorithm has a better repair effect, it effectively overcomes the sawtooth effect in Criminisi algorithm.
作者
李博文
白艳萍
LI Bo-wen;BAI Yan-ping(School of Science, North University of China, Taiyuan 030051, China)
出处
《数学的实践与认识》
北大核心
2018年第2期88-98,共11页
Mathematics in Practice and Theory
基金
国家自然科学基金(61774137)
山西省自然科学基金(201701D22111439,201701D221121)
山西省回国留学人员科研项目(2016-088)
关键词
图像修复
NSCT
领域方差
领域能量
image inpainting
nonsubsampled contourlet transform(NSCT)
neighbourhood variance
neighbourhood energy