中文摘要：

现有的基于压缩感知的超分辨率重建模型需要对高分辨率图像进行初始估计,而初始估计的准确与否直接影响图像重建的质量与迭代次数。针对此问题,本文引入非局部均值正则项以改进邻域嵌入方法,从而获得更加准确的高频初始估计;同时利用低分辨率图像的局部自相似性和多尺度结构相似性构建约束项,从而提出了一种基于改进的邻域嵌入和结构自相似性的超分辨率重建方法,充分结合两者的优势,增强了先验估计的表达能力。实验结果表明,相较于现有算法,本文提出的算法在客观评价指标和主观视觉质量上均有显著提高。

英文摘要：

Image super-resolution refers to the reconstruction process of a high resolution image from one or a set of low resolution images. The theory of compressed sensing plays an important role in super-res- olution reconstruction in recent years. However, existing super-resolution models based on compressed sensing tend to have an initial estimation for the high resolution image, while the quality of the estima- tion will directly affect the quality and efficiency of the reconstruction. In this paper,the image nonlocal similarity is introduced as a regularization term into the neighbor embedding method, which improves the quality of the initial estimation. We incorporate the image multi-scale similarity and the local similarity into the reconstruction model and propose an image super-resolution method based on neighbor embed- ding and structure self-similarity. The expression ability of prior information is enhanced because of com- bining the advantages of the two aspects. The experiment uses the peak signal to noise ratio （PSNR） and the structural similarity index measurement （SSIM） to compare the reconstructed image with the o- riginal image. Experimental results show that the proposed approach outperforms some state-of-the-art super-resolution methods in both subjective and objective evaluation criteria, which is able to enhance high frequency information in details and has a gain of 0. 47-1.84 dB in PSNR on average.