#TITLE_ALTERNATIVE#

• Main Author: HADI (NIM 13203106), SAMEKTA
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/11373
• Institution: Institut Teknologi Bandung

Lipofuscin autofluorescence process on Retinal Pigment Epithelium (RPE) eye retina can be used for detecting Age Macular Degeneration (AMD), Best's and Stargardt's Fundus Flavimakulatus diseases. Retina autofluorescence images taken from fundus camera Zeiss FF 450 plus IR are having low contrast so they have little advantage to diagnose process. This final project is designing an image processing system that increasing contrast, reducing blur, so a medical doctor is able to do a diagnose procedure easier. The system consists of four main processes: preprocessing, smoothing, deblurring, and contrast improvement. Preprocessing consists of image size reducing and image channel changing to reduce memory and time processing need. Image size reducing uses bicubic method. Image colour channel is changed from Red Green Blue (RGB) to grayscale. Smoothing reduces noises on image such as white, pepper, and salt noise. Smoothing uses a pixel-wise adaptive Wiener method based on statistics estimated from a local neighborhood of each pixel. Deblurring process to reduce image blur uses blind deconvolution Self-Deconvolving Data Reconstruction Algorithm (SeDDaRA) with pseudoinverse filter. After deblurring process, image has to pass contrast improvement which uses stretching algorithm. The processed image from designed system has high contrast and low level of blur so that medical doctor can do diagnose procedure easier.