计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (13): 187-190.
• 图形图像处理 • 上一篇 下一篇
康 凯，尹 东，张 荣
出版日期:
发布日期:
KANG Kai, YIN Dong, ZHANG Rong
Online:
Published:
摘要： 介绍了Criminisi算法，并针对其数据项可能导致虚假边缘出现的缺点，引入局部坐标系上的二阶偏导数对其进行改进。此外还对算法中最佳匹配块的搜索进行了优化。实验表明，该算法修复后的图像的主观效果要好于原来的算法，优化的搜索算法也使修复时间可观减少。
关键词: 图像修复, Criminisi算法, 局部坐标系, 差的平方和（SSD）
Abstract: This paper firstly introduces the Criminisi algorithm, and presents a new data term using the second order partial derivative on the local coordinate system regarding to the defects of the old data term which may lead to false edge. Besides, the search method of the best matching block is optimized. The experiments show the result images using this algorithm are better than the ones using the original algorithm in subjective perspective, and the repair time also has a considerable decrease because of the use of the optimized search algorithm.
Key words: image inpainting, Criminisi algorithm, local coordinate system, Sum of Squared Differences（SSD）
康 凯，尹 东，张 荣. 一种基于样图的图像修复改进算法[J]. 计算机工程与应用, 2014, 50(13): 187-190.
KANG Kai, YIN Dong, ZHANG Rong. Improved exemplar-based image inpainting algorithm[J]. Computer Engineering and Applications, 2014, 50(13): 187-190.
0 / 推荐
导出引用管理器 EndNote|Ris|BibTeX
链接本文: http://cea.ceaj.org/CN/
http://cea.ceaj.org/CN/Y2014/V50/I13/187
