RESOLVE Machine Paradigm 
Applied to Image Compression Algorithms

Joseph E. Hollingsworth
Computer Science Department
Indiana University Southeast
4201 Grant Line Road
New Albany, IN  47150  USA

jholly@ius.edu
Phone: +1 812 941 2425
Fax: +1 812 941 2637
URL: http://homepages.ius.edu/jholly
W. Christopher Lang
Mathematics Department
Indiana University Southeast
4201 Grant Line Road
New Albany, IN  47150  USA

wclang@ius.edu
Phone: +1 812 941 2391
Fax: +1 812 941 2637
URL: http://homepages.ius.edu/wclang

Abstract

Image compression algorithms such as JPEG are widely used when delivering internet content.  The algorithms typically have distinct phases and tradeoffs of quality versus performance.  It seemed to us that attempting to recast these algorithms as a machine would provide for an interesting challenge for the RESOLVE  machine paradigm.

Keywords

machine paradigm, image compression algorithms

Paper Category: position paper
Emphasis: research 

1.  Introduction

JPEG2000, the latest JPEG standard for image compression described in [Taubman02] contains many phases including: image pre-processing, image partitioning, applying a mathematical transformation to each partition, scaling the transformed data, and coding the scaled data.  These image compression algorithms appear to be good candidates for recasting as machines as described by the RESOLVE machine paradigm.  The authors of [Bucci02] outline several desirable properties of software machines including: hide data representations, hide algorithms used, hide when specific actions are executed, etc.  We seek to build an image compression machine possessing these properties.

2.  The Position

It is our position that we can attack this problem at several different levels including:

3.  Related Work

We base our work on [Bucci02] and [Weide94], in conjunction with our own experience at implementing sorting machine in various languages and with various algorithms.

4.  Conclusion

Image compression algorithms (e.g., JPEG) comprise many steps and possess opportunities for making tradeoffs in image quality versus performance.  These properties of image compression algorithms appear similar in nature to the properties possessed by sorting algorithms which have been successfully recast, and effectively captured as a sorting machine.  We understand that image compression algorithms may be quite difficult to recast, we also know that if we persist, we will learn a lot along the way.

At the workshop we would like to discuss with other participants the work that they might have done with image compression algorithms.  We would also gain from experiences related to recasting other algorithms as machines. 

References

[Bucci02]
Bucci, P., Heym, W., Long, T.J., Weide, B.W., “Algorithms and Object-Oriented Programming: Bridging the Gap” in 33rd SIGCSE: Technical Symposium on Computer Science Education, Covington, KY, February, 2002.
[Taubman02]
Taubman, Marcellin, JPEG2000: Image Compression Fundamentals, Standards and Practice, Kluwer, 2002.
[Weide94]
Weide, B.W., Ogden, W., and Sitaraman, M., “Recasting algorithms to Encourage Reuse”, IEEE Computer Society, 11(5), September 1994, pp. 80-88.