Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/4462
Title: Bi-orthogonal wavelets for compression, denoising and stereo matching of satellite images
Authors: Ji, Yanwen
Keywords: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
Issue Date: 2000
Abstract: Firstly, this thesis provides some background knowledge on wavelet theory. Bi-orthogonal wavelet filters designed by Yu and Ho [1] are introduced and applied in project experiments. Secondly, satellite image compression and satellite image denoising using bi-orthogonal wavelets are investigated. The simulation is performed on the 512x512 SPOT images. In image compression, wavelet transform is first performed on the image, adaptive differential pulse code modulation (ADPCM) and vector quantization (VQ) are used to encode the coarse frequency components and detail frequency components, respectively. Through this combination a compression ratio of 150:1 is obtained while maintaining a high peak signal noise ratio (PSNR) of 25.71dB on the SPOT image Airport. In image denoising by bi-orthogonal wavelets, the noisy SPOT image Airport with a PSNR of 18.23dB before denoising is improved to a PSNR of 23.45dB after denoising. Comparison results of our methods with existing method are presented. After obtaining the coordinates of matched points from a pair of Hongkong terrain stereo images, the terrain height can be estimated through a 3-D model. Stereo vision image is presented. Results show that MRHM is time efficient without decreasing the matched accuracy.
URI: http://hdl.handle.net/10356/4462
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:EEE Theses

Files in This Item:
File Description SizeFormat 
EEE-THESES_494.pdf
  Restricted Access
19.56 MBAdobe PDFView/Open

Page view(s) 50

385
checked on Sep 28, 2020

Download(s) 50

3
checked on Sep 28, 2020

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.