Author: Wong, Wing-man
Title: Vector quantization using quadtree and interpolative techniques
Degree: M.Sc.
Year: 1999
Subject: Image processing -- Digital techniques
Data compression (Telecommunication)
Coding theory
Hong Kong Polytechnic University -- Dissertations
Department: Multi-disciplinary Studies
Department of Electronic and Information Engineering
Pages: vii, 74, [5] leaves : ill. ; 30 cm
Language: English
Abstract: This thesis presents a review of vector quantization techniques used for encoding digital images. First the concept of vector quantization is introduced, then its application to digital image is explained. Mean-Residual, transform and interpolative vector quantizations are reviewed. Finally a new image compression scheme is proposed. This technique is based on mean-residual vector quantization (MRVQ) and block orientation so it is called block orientation vector quantization (BOVQ). The BOVQ scheme offers an excellent performance visually to the extent that the reproduced edges are not staircased and there is no noticeable blockiness in the reproduced image. This scheme can achieve bit rates of 0.3-0.8 bpp with moderate computational complexity. This scheme lies in encoding each input block by two-byte code vectors. One byte corresponds to the index representing the code vector which represents a tree structure as well. The other byte is used to represent the state of orientation of each block and its mean intensity. Variable block-size quadtree vector quantization is a method of image coding that combines quadtree vector quantization coding and clustering technique by which code vectors are generated to represent the blocks of an image. The new image compression scheme is investigated to reduce the required bit rate and to improve the reproduced image. Techniques used include variable block-size quadtree vector quantization, mean-residual vector quantization (MRVQ), transformation and interpolative vector quantization (IVQ). Variable block-size vector quantization VBVQ is adopted in the implementation of the quadtree vector quantization method. Therefore the image to be compressed is segmented into blocks of different size which depends upon a pre-defined value (threshold) of the degradation. The maximum block size used in the segmentation process for the proposed scheme is 16 x 16. For high-detailed regions, the block size will be reduced into four 8 x 8 pixel blocks. Any of these 8 x 8 blocks can be further reduced into four 4 x 4 pixel blocks. The bit rate for an encoded gray images with high quality can be below 0.5 bbp. The BOVQ coding scheme provides a more consistent image quality throughout all regions of the processed image. The schemes use an universal codebook which stores the required code vector. However, the method does not require side information because the index of a code vector also represents the corresponding tree structure. According to the simulation results, the BOVQ scheme outperforms the ordinary VQ in terms of quality by 2 to 4 dB. It can significantly reduce the bit rate compared with other vector quantization schemes, such as MRVQ.
Rights: All rights reserved
Access: restricted access

Files in This Item:
File Description SizeFormat 
b14719721.pdfFor All Users (off-campus access for PolyU Staff & Students only)3.44 MBAdobe PDFView/Open

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.

Show full item record

Please use this identifier to cite or link to this item: