Super-resolution imaging and its applications

Abstract

Super Resolution is an image processing technique that aims to generate high resolution (HR) images by utilizing a sequence of sub-pixel shifted low resolution (LR) images. During the past twenty years, most of the SR techniques have been applied on regular images taken with a hand-held camera. However, application of SR in the field of biomedicine is a state-of-the-art area for research. In fact, SR has not yet been applied in a Photo-thermal Imaging (PTI) system which provides a resolution exceeding the diffraction limit.

Therefore, in this project, a PTI system has been built up to obtain the Photo-thermal images of Red Polystyrene Beads and Red Blood Cells. Since the existing equipment has encountered some undesired problems during the experiment, a new CCD camera and a piezoelectric stage have been embedded into the PTI system. And the MATLAB programs have been modified accordingly to make the new equipment compatible with the system. By moving the highly sensitive piezoelectric stage with an accuracy of 0.1nm, several sets of sub-pixel shifted PTI images could be obtained.

Then SR has been implemented on these acquired LR PTI images. It has been found that SR techniques were not as straightforward for microscopic images compared to normal images, and many limitations need to be overcome to get good results. Therefore three different methods of SR were applied on these LR images for comparison. By adopting the optimum method, each sequence of LR images was able to generate a high quality HR image. SR shows significant improvement in the processed images as compared to the original PTI images, mainly in terms of resolution, sharpness and contrast.