Deblurring of Optical Images Due to Atmospheric Turbulence Effects Using Image Processing

As the field of optics grapples with its limitations in tackling remote sensing issues, deblurring images has emerged as a formidable challenge, particularly for initiatives undertaken by the U.S. Air Force. Our endeavor to delve deeper into this issue led us to curate a comprehensive literature review, thereby illuminating state-of-the-art procedures for image deblurring. In this study, we aim to address this conundrum by applying various deconvolution methods to denoise and deblur images marred by aberrations stemming from atmospheric turbulence effects. The distinct techniques employed for this purpose include the Wiener Filter, Regularized Filter, Lucy-Richardson Algorithm, Blind Deconvolution Algorithm, and Graph-Based Blind Image Deblurring. Our research has unveiled the effectiveness of specific deconvolution techniques in various situations, and highlighted how coding differences can potentially spawn distinct challenges. While image processing has proven to be a robust tool for denoising and deblurring, we have also noted that state-of-the-art deep learning methods, such as transformers, can yield superior performance. Nevertheless, deep learning's primary limitation lies in the acquisition of a comprehensive and diverse dataset, a prerequisite that often necessitates higher performance standards, thereby increasing time consumption and inefficiency. As a result, for tackling these ill-posed, inverse, and non-convex problems, we believe image processing offers a compelling solution, thanks to its speed, efficiency, and lesser computational resource requirements.