中文摘要：

超分辨率重建技术已经在相当多的领域中被应用,并且已经引起计算机视觉,图形图像处理等领域研究者的关注,目前是国内外图像处理领域的研究热点.随着对各种退化模型的研究,超分辨率重建越来越多的集中在空间域中.着重研究了基于变焦图像序列的超分辨率重建算法.首先,提出变焦变换的退化模型.然后从LR（低分辨率）图像序列中选择一副图像,将其插值放大后的图像作为初始HR图像.接着,根据退化模型和计算出的特征匹配点对,计算出某一幅LR图像和HR图像之间的变换公式.通过该变换公式,将HR（高分辨率）图像变换为模拟LR图像.求得模拟LR图像与这幅实际LR图像之间的差值.当差值大于设定阈值时,则将差值反投影并叠加至HR图像上,由此更新HR图像.最后,分别对仿真变焦图像序列、实际变焦图像序列进行超分辨率重建.对重建后的结果,利用峰值信噪比、结构相似度和高频能量在总能量中所占比例这三种评判标准,从三个角度进行评价.最终,实验结果显示,重建方法对变焦图像序列能取得较好的重建效果.

英文摘要：

The purpose of super resolution（SR）is to reconstruct a high-resolution（HR）image with a higher quality version effect from one or a sequence of low-resolution（LR）images,which are captured from the same scene and subsampled as well as shifted with sub-pixel precision.The resolution enhancement technique of super resolution has been widely applied recently,for example,the computer version and image processing.With the research of image degradation model,reconstructing images in the spatial domain become more and more popular.In this thesis we focus on super-resolution reconstruction methods of zooming image sequence.However,these methods do not pay enough attention to the situation where the size of target in the zooming image sequence changes greatly.In this thesis we focus on super-resolution reconstruction methods of zooming image sequence.Firstly,we presented zooming degradation model based on the low-resolution image sequence.A target-size image produced from LR image by interpolation is then selected as the initial high-resolution image.Based on the degradation model and the matching points generated by scale-invariant feature transform（SIFT）respectively in HR and LR image,the transform formula is pro-posed.By this transform formula,a simulated LR image is generated from the HR image.Then we computed the difference between the simulated LR image and the real observed LR image mentioned in computing transform formula.When the difference is higher than some threshold,the difference is added into the HR image by back projection so that the HR image is updated.Finally,the improved algorithm is employed on simulated zooming LR image sequence and real zooming LR image sequence.Peak signal to noise ratio（PSNR）,structural similarity index（SSIM）and energy ratio between high-frequency components and the whole image are calculated to evaluate the quality of our rebuilt HR image.The experimental results showed that the HR image reconstructed by our proposed SR algorithm with zooming LR image sequen