A detailed analysis of different CNN implementations for a real-time image processing system

Publikationstyp
Conference Paper
Date Issued
2000
Sprache
English
Author(s)
Wiehler, Klaus  
Perezowsky, Michael 
Grigat, Rolf-Rainer  
Institut
Bildverarbeitungssysteme E-2  
TORE-URI
http://hdl.handle.net/11420/10202
Start Page
351
End Page
356
Citation
Proceedings of the 2000 Sixth IEEE International Workshop on Cellular Neural Networks and their Applications (CNNA 2000): 351-356 (2000)
Contribution to Conference
6th IEEE International Workshop on Cellular Neural Networks and Their Applications Proceedings (CNNA 2000)  
Publisher DOI
10.1109/CNNA.2000.877354
Scopus ID
2-s2.0-0033697761
Publisher
IEEE
A detailed analysis for different implementations of a real-time CNN signal processing systems is presented. The algorithm for signal reconstruction has been realized in hardware (analog VLSI, multi-FPGA system) and in software (TriMedia VLIW, Intel Pentium processor). All implementations are fully functional and embedded in a system environment. Due to the high computational complexity which is needed to solve the nonlinear CNN-equations and the requirements which are different for each application, an efficient implementation has to be tailor-made. In this publication we analyze different realized implementations regarding prototypical prerequisites.
DDC Class
004: Informatik