Date of Award
January 2020
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Computer Science
First Advisor
Ronald A. Marsh
Abstract
The goal of this research was to develop a neural network that will produce considerable improvement in the quality of JPEG compressed images, irrespective of compression level present in the images. In order to develop a computationally efficient algorithm for reducing blocky and Gibbs oscillation artifacts from JPEG compressed images, we integrated artificial intelligence to remove blocky and Gibbs oscillation artifacts. In this approach, alpha blend filter [7] was used to post process JPEG compressed images to reduce noise and artifacts without losing image details. Here alpha blending was controlled by a limit factor that considers the amount of compression present, and any local information derived from Prewitt filter application in the input JPEG image. The outcome of modified alpha blend was improved by a trained neural network and compared with various other published works [7][9][11][14][20][23][30][32][33][35][37] where authors used post compression filtering methods.
Recommended Citation
Amin, Md Nurul, "Removal Of Blocking Artifacts From JPEG-Compressed Images Using Neural Network" (2020). Theses and Dissertations. 3089.
https://commons.und.edu/theses/3089