Image restoration techniques for removal of blurred images / Ameer Naeemi Zyarah

A digital image is a two-dimensional numerical array that is produced to record a faithful yet significant scene, however more often than not the recorded image invariably represents a blurred version of the original scene. Blurring is introduced in the process of imaging due to relative motion between camera and scene, atmospheric turbulence, etc. Hence, image restoration is a fundamental research topic in the realm of image to obtain an optimal estimate of the original image given the degraded image. This research project explores the motion blur which arises from the relative motion between camera and scene. In this study, four different techniques are used to remove the motion blur. They are Direct Inverse filter, Wiener filter, Constrained Least Squares filter, and Lucy Richardson algorithm, to restore degraded image (motion blurred image). In this research project, an original image is motion blurred at fixed length (30 pixels) along with different angles (θ). These degraded images are then restored with the derived image restoration techniques. Statistical error image metrics (MSE and PSNR) and Human Visual System feature-based metric (SSIM) are then computed to evaluate and analyze the quality of the restored images using the aforementioned image restoration techniques. Experimental and simulation results show that Wiener filter is the best-performing image restoration technique, followed by Direct Inverse filter, Constrained Least Squares, and lastly, Lucy Richardson algorithm