The work of the thesis is based on an investigation of the potential for utilising the Integrated Services Digital Network (ISDN) for image transmission within a medical scenario. The work initially identifies the major requirements of the application, and suggests that some form of compression of the data is necessary, with the possibility of progressive enhancement in some situations.
Details
- Author: Neal Snooke
- Supervisor: Mr. David Price, UWA
- Thanks to: Dr. Horst Holstein, and Prof. Mark Lee UWA.
- Submitted: Sept 1993
Outline
We consider a number of state of the art encoding approaches and provide evaluation of each within the mdeical imaging context. The approach is taken that the many advantages of lossy encoding often outweigh the disadvantages, although effective use of these techniques can only be made when characteristics of the data redundant to the future use of the image is identified. Sources of redundancy are located at several levels, not only visual sensitivity, but from statistical and spatial structure, based on external knowledge about the image and application.
A Wavelet transformation was selected due to a number of very useful characteristics which we exploit in the development of a compression scheme capable of supporting both the statistical and human visual models as well as providing a framework for allowing higher level regional information to be used to allow non uniform quality selection. Thus the potential ability to preserve the diagnostic information content of the image which can be lost with general lossy encoding techniques is sought, whilst still maintaining the high compression rates characteristic of lossy compression.
The scheme can easily be developed into a high ratio archive algorithm; the implications of this use are also considered throughout the work.