深度学习是近年来最流行的科研潮流之一，广泛应用于多个领域。传统的遥感图像重建方法具有许多局限性，再加上光学器件的限制，使得获取图像中的细节信息变得十分困难。这也使得深度学习在图像超分辨率重建中的运用意义变得格外重要。
本文介绍了目前各种深度学习方法，研究了利用深度学习进行图像超分辨率重建的可能。本文选取了SRGAN与DDPM两种方法，介绍了二者的基本原理与实现过程，进行遥感图像的超分辨率重建，得到处理结果，并计算图像的质量客观评估指标。
为了验证方法的有效性，将本文所采用的两种方法与插值法和SRCNN两种经典方法进行对比评估。实验结果表明，本文使用的方法重建出的图像纹理更加细致，图像真实感更强，放大后的细节更加丰富，相对经典方法有更进一步的提升。

Deep learning is one of the most popular scientific research trends in recent years, with extensive applications in many fields. Traditional remote sensing image reconstruction methods have many limitations, coupled with the limitations of optical devices, making it very difficult to obtain detailed information in the image. This also makes the application significance of deep learning in image super-resolution reconstruction particularly important.
This article provides an overview of various current deep learning methods and investigates the possibility of using deep learning for image super-resolution reconstruction. This article selects two methods, SRGAN and DDPM, introduces their basic principles and implementation processes, performs super-resolution reconstruction of remote sensing images, obtains processing results, and calculates objective evaluation indicators for image quality.
In order to verify the effectiveness of the method, the two methods used in this article will be compared and evaluated with interpolation and SRCNN, two classic methods. The experimental results show that the method used in this article reconstructs images with finer textures, stronger image realism, and richer details after magnification, which is further improved compared to classical methods.