Blind Deconvolution Method of Image Deblurring Using Convergence of Variance

摘要

Images are used for both aerial and space imagery applications, including target detection and tracking. The current problem concerning objects in geosynchronous orbit is that they are dim and hard to resolve because of their distance. This work will further the combined effort of AFIT and AFRL to provide enhanced space situational awareness (SSA) and space surveillance. SSA is critical in a time when many countries possess the technology to put satellites into orbit. Enhanced imaging technology improves the Air Force's ability to see if foreign satellites or other space hardware are operating in the vicinity of our own assets at geosynchronous orbit. Image deblurring or denoising is a crucial part of restoring images that have been distorted either by movement during the capture process, using out-of-focus optics, or atmospheric turbulence. The goal of this work is to develop a new blind deconvolution method for imaging objects at geosynchronous orbit. It will feature an expectation maximization (EM) approach to iteratively deblur an image while using the convergence of the image's variance as the stopping criteria.