Evaluation and optimization of central vision compensation techniques

Abstract

Non-costly, non-invasive, safe, and reliable electronic vision enhancement systems (EVES) and their methods have presented a huge medical and industrial demand in the early 21 st century. Two unique, vision compensation and enhancement algorithms are reviewed and compared, qualitatively optimizing the view of a restricted (or truncated) image. The first is described as the convex or "fish-eye" technique, and the second is the "cartoon superimposition" or "Peli" technique (after the leading author for this research). The novelty in this dissertation is in presenting and analyzing both of these with a comparison to a novel technique, motivated by characterization of quality vision parameters (or the distribution of photoreceptors in the eye), in an attempt to account for and compensate reported viewing difficulties and low image quality measures associated with these two existing methods. This "partial cartoon" technique is based on introducing the invisible image to the immediate left and right of the truncated image as a superimposed cartoon into respective sides of the truncated image, yet only on a partial basis as not to distract the central view of the image. It is generated and evaluated using Matlab ® to warp sample grayscale images according to predefined parameters such as warping method, cartoon and other warping parameters, different grayscale values, as well as comparing both the static and movie modes. Warped images are quantitatively compared by evaluating the Root-Mean-Square Error (RMSE) and the Universal Image Quality Index (UIQI), both representing image distortion and quality measures of warped, as compared to original images for five different scenes; landscape, close-up, obstacle, text, and home (or low-illumination) views. Remapped images are also evaluated through surveys performed on 115 subjects, where improvement is assessed using measures of image detail and distortion. I would like to express my appreciation to several groups of people on the academic, institutional, and personal levels. I would like to first, and foremost, show my deepest gratitude to my advisor and mentor Dr. Smith for believing in my abilities and never giving up hope; I wish you always the best, and most of all guidance. I would also like to thank Dr. Odom, who has been a great source of motivation, ideas, as well as professional and moral support and encouragement. Dr. Means, Dr. Thompson, and Dr. Huebsch are appreciated for serving on the AEC and having generously given their time and expertise to raise this work to a higher professional level.