Massively parallel quantization implementation using simulated annealing

Engelen, L. van (2006) Massively parallel quantization implementation using simulated annealing.

[img]
Preview
PDF
729kB
Abstract:This thesis treats the mapping of a quantization algorithm on the Linedancer parallel processor architecture which was done at Oce. This algorithm is based on Simulated Annealing and uses a Markov Random Field image model. The mapping of the algorithm was supported by the use of the Evolutionary Design methodology. The mapping of the algorithm was nished successfully, although the output quality was worse than that of established algorithms. Throughout development an executable prototype was used to guard the quality of the mapping. We identfied four typical subphases in the top down part of the Evolutionary Design methodology. The implementation of the quantization algorithm benets from the implementation on the Linedancer, demonstrated by a higher processing speed when compared with a Pentium. Finally some directions for additional research are suggested.
Item Type:Essay (Master)
Clients:
Oce
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science MSc (60300)
Link to this item:http://purl.utwente.nl/essays/57349
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page