Probabilistic Methods for Image Generation and Encoding.
Abstract
The research described herein involves a new digital image compression algorithm. A hybrid low-pass filter/fractal technique was developed, taking advantage of the fact that fractals afford a remarkable way of compressing high frequency data. The low frequency content of the image is stored without loss, and then fractal image processing is used to touch-up the high frequency features, with some loss. An image compression software system has been built around this. The system handles 24-bit digital images of arbitrary resolution. Compression is quick - for example, the system compresses a still 1280 x 1024 (high-definition) image in a few seconds on a Personal Iris (6 MFlops); and also decompresses in a few seconds.. At low compression (around 15:1) the compression is visually lossless. At high compression (around 100:1) there is some loss, but not much. Digital image compression, Fractals, Iterated function systems.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 15, 1993
- Accession Number
- ADA274294
Entities
People
- Marc A. Berger
Organizations
- Georgia Tech