|||
ROF Model-based Method For Wavelet
Image Inpainting in Pixel Domain
摘 要 图像修补问题是图像处理领域的重要内容之一,本文简要介绍了图像修补问题及现在一些主要的图像修补技术。根据图像修补的目的,提出小波图像修补问题不必局限于小波域求解,也可以在空域得到解决的思想。文中给出了该思想的具体实现依据和步骤,并提出一种在空域中基于ROF(Rudin-Osher-Fatemi)模型来实现上述思想的小波图像修补方法。图像修补仿真实验的结果表明,本文提出的方法可以得到较好的图像修补视觉效果,峰值信噪比(Peak signal to noise ratio, PNSR)小于10dB的缺损图像经处理后其PSNR达到约30dB,较以前的小波图像修补方法,该算法更加显著地提高缺损图像的PSNR。因而,本文在空域实现了小波图像修补,本文提出的思想是可行的。
关键词 图像处理;图像修补;ROF模型;小波
Abstract Image inpainting is one of the most important problems in image procesing.
Image inpainting and some of its main techniques are presented in this paper
briefly. According to the target of image inpainting, an idea to solve wavelet
image inpainting not only in wavelet domain but also in pixel domain and its reason
and steps to be implemented are given. A wavelet image inpainting method in pixel
domain based on ROF (Rudin-Osher-Fatemi) model is proposed to put the above
idea into practice. Experiments showed that the method could get good visual effect
of image inpainting, with higher PSNR compared with former methods——PSNR
of damaged image increased from less than 10dB to 30dB after processing using our proposed method.
Therefore, wavelet image inpainting is achieved in pixel domain, which means that our proposed thought
is practicable.
Keywords image processing; image inpainting; Rudin-Osher-Fatemi model; wavelet
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-12-21 20:30
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社