This PhD project aims to develop a deep learning based method to improve the resolution of images and videos based on public dataset and experimental dataset. High-resolution (HR) images and videos are strongly demanded in many applications, such as satellite imaging, medical diagnostic, forensic imaging and video surveillance systems, not only for offering better visualisation but also for extracting additional details. However, in many cases it is difficult to obtain the demanded HR images/videos due to the high cost and inherent physical constraints of the high precision optics and sensors, or the limitation of bandwidth of data communication. Super-resolution techniques have emerged as an alternative low-cost solution, which aims to produce a single HR image or a set of HR images from a sequence of observed low-resolution images captured from the same scene.