University of Houston

When the input image is high-resolution, more information can be retrieved for better post processing such as detection, recognition, segmentation, identification, and visualization. The need for high-resolution images occurs in health care, where the doctor needs a high-quality image of the patient in order to make better decisions or perform surgery. For example, breath-holding high-speed MRI acquisition is challenging, if the data collected in a short amount of time might be insufficient for the diagnosis of patients, due to the low quality of the acquired frames. A major challenge in cardiac MRI is the motion of the heart itself and the motion of the diaphragm in the chest, which is moving the heart up and down when the patient breathes. Therefore, artifacts arise due to the changes in signal intensity, resulting in blurry images. Assuming that we can collect multiple frames of low quality MRI images over time, without a breath-hold, we can reconstruct a better quality image by fusing them. Some surveillance cameras have low acquisition speed and collect low quality images, due to the storage space restrictions or limited network bandwidth to transfer the data. Thus, in many surveillance scenarios, we obtain only a few low quality frames of an object or a subject.

The goal of this dissertation is to study previous approaches related to image quality, find their limitations, and introduce new approaches to solve them. Since it is difficult to build an algorithm that will work for all cases, such as MRI images and surveillance cameras, we divide the problem into sub-problems and thoroughly explain our approaches how to solve each of the sub-problems.