Document Details

Document Type:   Thesis
Title:   A Comparative Quantitative Approach to Digital Image Compression
Author:   Michael Wyllie
College:   Science
Degree Program:   Physical Science, M.S.
Degree:   Master of Science
Committee Director:   Dr. Ralph Oberly
Document Availability:   Document available for World-Wide access.
Date of Defense:   11/22/2006

Image compression has become a standard procedure when sharing files over the Internet. Lack of bandwidth forced early users to devise ways of compressing image and data files to manageable sizes. Image files were a particular problem due to the large amount of information contained within them. Compression schemes fall into two basic types, lossless and lossy. Lossless compression enables all of the data contained in the original to be reproduced. Lossy compression discards “unneeded” data using certain techniques to reduce file size. Lossy compression schemes gained great popularity due to their ability to produce smaller files compared to lossless compression. One way of comparing the effects of image compression algorithms is to look at changes in their histograms and statistical parameters such as mean, standard deviation and median values. By using such programs as Adobe Photoshop and ER Mapper it is possible to generate these values and compare them to the original image as compression increases. Changes in these values may also provide an objective way to determine if the image is of sufficient quality to extract useful data. Image subtraction is another powerful tool in the analysis of the changes due to compression. This technique shows on a per pixel basis the differences caused by the algorithm being employed as compared to the original or another compressed image. This paper will compare the effects of compression algorithms. These use advanced techniques to achieve their objectives. The compression methods are JPEG 2000 and ECW. Both use wavelet transforms to achieve a reduction in file size. JPEG 2000 is an open standard developed by the Joint Photographic Experts Group to replace the current widely used JPEG algorithm. ECW (Enhanced Compression Wavelet) is a proprietary method developed by Earth Resources Mapping. This paper will also compare how the same algorithm as applied by two companies can yield differing results. Images and statistics will show the variations in the implementation of JPEG 2000 by programs from Algo Vision LuraTech and Earth Resources Mapping. These results will also be compared to the ECW compression method to obtain the maximum amount of data at a given compression ratio from an image.  

Download Document:   wyllie-michael-2006-ma.pdf