Wavelet based image compression using FPGAs [Elektronische Ressource] / von Jörg Ritter

Wavelet based image compressionusing FPGAsDissertationzur Erlangung des akademischen Gradesdoctor rerum naturalium (Dr. rer. nat.)vorgelegt derMathematisch Naturwissenschaftlich Technischen Fakultätder Martin Luther Universität Halle Wittenbergvon Herrn Jörg Rittergeb. am 12. April 1971 in GreizGutachter:1. Prof. Dr. Paul Molitor, Martin Luther University Halle Wittenberg, Germany2. Prof. Dr. Scott Hauck, University of Washington, Seattle, WA, USA3. Prof. Dr. Thomas Rauber, University of Bayreuth, GermanyHalle (Saale), 06.12.2002urn:nbn:de:gbv:3-000004615[ http://nbn-resolving.de/urn/resolver.pl?urn=nbn%3Ade%3Agbv%3A3-000004615 ]AbstractIn this work we have studied well known state of the art image compression algorithms. These codecs arebased on wavelet transforms in most cases. Their compression efficiency is widely acknowledged. Thenew upcoming JPEG2000 standard, e.g., will be based on wavelet transforms too. However, hardwareimplementations of such high performance image compressors are non trivial. In particular, the on chipmemory requirements and the data transfer volume to external memory banks are tremendous.We suggest a solution which minimizes the communication time and volume to external random accessmemory. With negligible internal memory requirements this bottleneck can be avoided using the partitionedapproach to wavelet transform images proposed in this thesis.