摘要： 针对单帧低分辨率图像的超分辨率重建问题，提出一种改进的小波局部适应插值的超分辨率重建方法，该方法能够弥补重建图像边缘不平滑的缺陷。结合小波变换与可分离高低频信息的特性，提出一种综合两者优点的单帧图像超分辨率重建算法。实验结果表明，采用该算法得到的重建图像不仅能较好地保留原始图像的细节信息，提高图像的空间分辨率，并能提高图像的峰值信噪比，更适合人眼视觉系统。

关键词: 超分辨率, 小波变换, 局部适应插值, 最大相关性, 图像重建

Abstract: Aiming at the problems in Super-Resolution(SR) reconstruction of single Low-Resolution(LR) image, an improved local-adaptation interpolation method based on wavelet transform for image SR reconstruction is proposed. The method makes up for the shortcomings of that the edge of the reconstruction image is not smooth enough, and it maintains the edge and details of images. Combined with the characteristics of that wavelet transform can separate high and low frequency information. The single image SR reconstruction algorithm combined with the advantages of them is presented. Experimental results show the reconstructed image using this method can not only maintain the details information of original image very well, improve the image spatial resolution, but also improve the Peak Signal Noise Ratio(PSNR) of image. It is more suitable for human visual system.

Key words: Super-Resolution(SR), wavelet transform, local-adaptation interpolation, max-relativity, image reconstruction

中图分类号: 

• N945