摘要
近年来有关图像复原的研究成果大量涌现,焦点大多集中在图像中的纹理合成或面向结构的图像润饰。分析的几个重要算法,即基于同步的结构纹理填充算法、基于统计的图像润饰算法和基于样本的图像润饰算法,突破了以往研究的局限性,取得了较好的复原效果。基于同步的结构纹理填充算法,首先将纹理合成和图像润饰技术同时应用于图像恢复,基于统计的图像润饰算法则全面利用了图像的全局信息,基于样本的图像润饰算法则将纹理合成和图像润饰技术的特性融于一体。对此进行了分析探讨,并在此基础上总结了目前图像复原研究状况。
The research on image restoration has been made much in recent years. The opinions are much more focused on texture synthesis or inpainting of the image. The algorithms that we researched, have been beyond the limitations and make good results, which are synchronization-based filling ofstructure and texture, statistic-basedinpaintingandexemplar-basedinpainting. Thealgorithnofsynchronization-based filling of structure and texture first makes use of texture synthesis and inpainting in image restoration. The algorithm of statistics-based inpainting utilizes whole information in inpaiting. The algorithm of example-based inpainting mixes the characteristics of texture synthesis and inpaiting. These were analyzed. At last current status was summed up.
出处
《计算机工程与设计》
CSCD
北大核心
2006年第2期184-187,共4页
Computer Engineering and Design
基金
浙江省自然科学基金项目(M603228)
关键词
图像复原
纹理合成
图像润饰
基于同步的结构纹理填充
基于统计的图像润饰
基于样本的图像润饰
image restoration
texture synthesis
inpainting
synchronization-based filling of structure and texture
statistics-based inpainting
exemplar-based inpainting
