Date of Award


Document Type

Honors Thesis


Computer Science

First Advisor

John MacCormick




In this project, we investigated the effectiveness of the Super-Resolution algorithm on a distant whiteboard scenario. Imagine a person taking a video or an image sequence of a whiteboard from a distance. Due to the limitation of camera resolution and the distance, the words in the whiteboard images are sometimes illegible. Though there exist applications to enhance the image quality, the resolution limit caused by the distance is difficult to overcome. Super-resolution, a class of techniques in the field of Computer Vision, enables us to enhance the quality of the images by utilizing the information of multiple low quality images and fuse them to obtain an higher quality image. In this project, we study spatial domain SR algorithms, experiment on an implementation of it in OpenCV, and try to apply it to our distant whiteboard data set. Due to the complexity of the OpenCV implementation, we first perform black-box empirical analysis on the running time of the algorithm. Quantitative analysis on the quality of the resulting images are then conducted to understand the effect of different parameter values for the algorithm. Finally, we attempt to improve the quality of the output by adjusting the algorithm to test the whiteboard scenario and experimenting on different types of optical ow algorithms.