The JPEG Still Picture Compression Standard

Gregory K. Wallace, CACM 34(4), April 1991

Summary | Relevance | Rating | Back to Index

Content Summary

Description of the (then-immature) JPEG standard, which is actually a "toolkit" of various compression techniques.

JPEG goals:

State of the art
General-purpose
computationally tractable
allow lossy (variable-quality), lossless, and hierarchical encoding
allow progressive encoding (based on image features rather than sequential scan)

Some components, such as the DCT implementation and the encoding of image meta-information (aspect ratio, color space, etc.), are omitted from the standard. The standard allows for multiple images with separate quantization tables and specifies an interchange format. (How is colormap information handled? The article wasn't clear on this.)

Predictive lossy or lossless coding provides the functionality for hierarchical (multiresolution) coding: downsample, interpolate, upsample, predict, difference.

Relevance to Multimedia

A general-purpose, portable set of standards for image coding, designed explicitly to deal with photograph-quality (as opposed to halftone etc.) images. This emphasis is what differentiates JPEG from other graphic compression standards and makes it particularly relevant to multimedia.

Rating

4 out of 5: discrete cosine transfers are boring, but at least this is a readable exposition and provides some history of how the JPEG committee came to be and how the standard was arrived at.

Armando Fox ([email protected])