Abstract: In many photoelectric imaging fields，the obtained image is commonly poor quality and low resolution，due to the factors of motion，blurring，down-sampling，noise and so on.In order to maintain the basic information of CCD image，improve image visual effect and spatial resolution，this paper proposes a new image interpolation algorithm based on the decomposition of Non-Subsampled Contourlet Transformation（NSCT）.In this novel interpolation algorithm，high frequency components of an image are extracted by NSCT，the original low resolution image zoomed with bilinear interpolation is served as the low frequency part in reconstructed image.Then，these two parts are combined with NSCT inverse transformation to produce reconstructed high resolution image.Experimental results show that the proposed algorithm correlates the global structure and protects more clear edge information of low resolution image，so as to reach better visual effect than traditional interpolation of zoomed image.

Key words: image zoom, interpolation, non-sample Contourlet translation, edge information

摘要： 由于成像系统内部和外部的因素，使得获取的图像产生退化。为了尽可能地保持CCD图像的基本信息，提高图像的视觉效果和空间分辨率，同时克服传统插值算法存在的缺陷，提出了一种基于非采样Contourlet变换（Non-Subsampled Contourlet Transform，NSCT）分解的图像插值算法，利用NSCT提取高频成份，原始低分辨率图像经过双线性插值放大充当低频部分，再经过NSCT逆变换得到放大后的高分辨率图像。实验结果表明，相对于传统的双线性插值算法，所提出的基于NSCT分解的图像插值算法考虑到全局相关性，得到了更加清晰的边缘信息，从而得到更好的图像插值放大后的视觉效果。

关键词: 图像放大, 插值, 非下采样Contourlet变换, 边缘信息