Wavelet transformation in image processing

Pao Yue-kong Library Electronic Theses Database

Wavelet transformation in image processing

 

Author: Chan, Sing-ping Edwin
Title: Wavelet transformation in image processing
Degree: M.Sc.
Year: 1997
Subject: Image compression -- Data processing
Video compression
Coding theory
Wavelets (Mathematics)
Hong Kong Polytechnic University -- Dissertations
Department: Multi-disciplinary Studies
Pages: vii, 126 leaves : ill. ; 30 cm
Language: English
InnoPac Record: http://library.polyu.edu.hk/record=b1250460
URI: http://theses.lib.polyu.edu.hk/handle/200/3097
Abstract: This thesis discusses one of the relatively recently developed image transformation techniques and its application to digital image and video compression. This transformation technique uses multiple resolutions of the signal and is based on the wavelet theory. In this thesis, we are trying to use the simplest wavelet two-bands filter bank coding scheme to implement the image decomposition and reconstruction. All the images used are 8-bit gray level 256x256 size standard images. They can be disintegrated down to six levels subband images which sizes are 128x128, 64x64, 32x32, 16x16, 8x8 and the last level is 4x4. As wavelet decomposition is multiresolution in nature, so the upper level details subbands including the finial approximation subband are coming from the previous approximation subband. A Lenna image decomposed at one, three as well as six levels can demonstrate this multiresolution theory. Both of our wavelet high-pass and low-pass filters in wavelet transform are originated from the filter coefficients which are determined by the inputted alpha and beta values. By applying difference alpha and beta values, we can get a series of wavelet filters. The Lenna image is decomposed and reconstructed by this set of filters and their Root Mean Square Error (RMSE) is computed to determine the filters performance. Some of filters show perfect reconstruction which means the RMSE is zero, whereas some of the extremely worst filters can generate the RMSE greater than 23. Since the error will be accumulated during transformation, so the higher the wavelet level, the greater the RMSE will get. In order for image compression, some of the unimportant information can be removed. Thus we also cut-off some details information or the approximation information at difference level to understand their error change. Besides the Lenna image, the baboon and pepper images are also tested. Their RMSE and rebuild images are printed for comparison. If more information is cut, the higher the RMSE will be gotten. Also the approximation information contain the major data for reconstruction, so it should not be reduced. Only the details information, especially for the higher wavelet level, could be sacrificed.

Files in this item

Files Size Format
b12504609.pdf 5.611Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

     

Quick Search

Browse

More Information