Iterated Transform Image Compression
Abstract
The objectives of this research were to present in a clear manner an algorithm based on iterated transforms that can be used to compress grayscale images and demonstrate the algorithm and present results for various images. The theoretical framework for iterated transform image compression has been generalized to include noncontractive transforms. The method has been used to encode the 512 x 512 8-bpp image of Lena at a compression of 15.9:1 with the decode image having a root mean square error of 6.33 (32.1 dB). Other images have been encoded at various encoding conditions with the resulting compressions ranging from 10:1 to 63:1. It was shown that the relaxation of the contractivity constraint can lead to improvement in the fidelity of decoded images.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1991
- Accession Number
- ADA233655
Entities
People
- E. W. Jacobs
- R. D. Boss
- Y. Fisher