基于小波变换与Bézier曲面插值相结合的图像插值方法  

Image interpolation by combining wavelet transform with Béizer surface

作　　者：吴伟仁[1] 闫磊[1] 田金文[1] 

机构地区：[1]华中科技大学图像处理与智能控制教育部重点实验室,湖北武汉430074

出　　处：《哈尔滨工业大学学报》2004年第11期1468-1470,1506,共4页Journal of Harbin Institute of Technology

基　　金：国家高技术研究发展计划资助项目(2001AA135020).

摘　　要：为提高遥感图像的分辨率,提出首先对图像进行小波分解,然后对小波分解后的相应高频子带进行B啨zier曲面插值,再通过小波逆变换后获得高分辨率图像的图像插值方法.实验结果表明:该方法较好地保持了图像中丰富的高频成分,插值后的图像具有很好的视觉效果和较高的信噪比,克服了传统的插值方法使图像高频部分损失、细节被模糊的缺点;该算法结构简单、易于快速实现,是一种行之有效的超分辨率算法.Abundance of high frequent components in image is preserved by combining wavelet transform and Béizer surface. The quality of image interpolation is good subjective visual effect and high ratio of signal and noise in objectivity. The image interpolated is of abundant detail. There is not obvious aberration in the interpolated image. The high frequent loss and interesting detail fuzzy had been overcome in that our method compare with some traditional methods. The algorithm structure is sample and fast implement is easy. The experiment results show that our method is an effective image super-resolution method.

关 键 词：超分辨率 图像处理 小波变换 BÉZIER曲面 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]