摘要
文章对纹理合成Criminisi算法中优先级的确定以及匹配块的选取分别作了改进,提出了一种基于改进优先级的加权匹配的图像修复算法,在优先级中考虑了图像显著结构的影响,使改进的优先级更加精确;提出了更为合理的加权匹配因子,并根据图像结构相似性特征对图像进行分块修复;实验表明,改进后的算法不仅克服了现有算法可能存在的偏差延续问题,使图像修复的结果更加符合人们的视觉效果,而且大大缩短了修复时间。
In this paper, a weighted matching image restoration algorithm based on the modified priori- ty is given by improving the priority setting and the matching block selection in the Criminisi texture systhesis algorithm. Firstly, the influence of the image significant structure is taken into account in the priority of the new algorithm, which makes the modified priority more accurate. Secondly, the matching block in the new method is given a more reasonable weighted matching factor. Finally, the image is restored by the new method based on the image segmentation according to the image structure similar characteristics. The experimental results show that the algorithm introduced in the paper a- voids the possible bias diffusion by the current image restoration algorithms, makes the restored image more consistent to human being＇s visual effects, and reduces greatly the time consumed.
出处
《合肥工业大学学报（自然科学版）》
CAS
CSCD
北大核心
2013年第1期113-118,共6页
Journal of Hefei University of Technology：Natural Science
基金
教育部科学技术研究重大资助项目(309017)
留学回国人员科研启动基金资助项目(2010JYLH0322)
安徽省自然科学基金资助项目(11040606M06)
国家大学生创新性实验计划资助项目(101035937)
关键词
纹理合成
图像修复
优先级
匹配块
分块
texture synthesis l image restoratiom priority l matching block~ image segmentation
