摘要
将基于样本的图像修补方法用于视频修复,利用视频序列时间上的相关性对修补方法进行改进,提出一种对视频序列损伤的修补方法,可解决检测不到的视频损伤的修补难题.实验表明,与基于样本的图像修复方法相比,文中方法的修补效果更好,避免了错误填充和错误蔓延,提高了修补效率.
This paper applies exemplar-based image inpainting to video restoration, and improves the repairing method by using time correlation characteristic of video sequences. The new method of video damage restoration can repair damages which are difficult to be detected in videos. Experiments show that, comparing with exemplar-based image inpainting, the method is more efficient in video restoration, can avoid filling errors and error spread, and has better efficiency.
出处
《应用科学学报》
EI
CAS
CSCD
北大核心
2010年第2期163-169,共7页
Journal of Applied Sciences
基金
浙江省科技计划项目基金(No.2007C21014)资助
关键词
视频修复
图像修补
优先权
最佳样本块
纹理综合
video restoration, image inpainting, priority , best exemplar patch, texture synthesis