Search
Abstract
In this paper we present a genetic algorithm (GA) for creating hypothetical virtual portraits of historical figures and other individuals whose facial appearance is unknown. Our algorithm uses existing portraits of random people from specific historical period and social background to evolve a set of face images potentially resembling the person whose image is to be found. We then use portraits of the person's relatives to judge which of the evolved images are most likely to resemble his/her actual appearance. Unlike typical GAs, our algorithm uses a new supervised form of textit{fitness function} which is affected by the evolution process. Additional description of requested facial features can be provided to further influence the final solution (i.e. the virtual portrait). We present an example of a virtual portrait created by our algorithm. Finally, performance of a parallel implementation developed for the KASKADA platform is presented and evaluated.
Authors (3)
Cite as
Full text
- Publication version
- Accepted or Published Version
- License
-
open in new tab
Keywords
Details
- Category:
- Articles
- Type:
- artykuły w czasopismach recenzowanych i innych wydawnictwach ciągłych
- Published in:
-
TASK Quarterly
no. 16,
pages 145 - 162,
ISSN: 1428-6394 - Language:
- English
- Publication year:
- 2012
- Bibliographic description:
- Krawczyk H., Proficz J., Ziółkowski T.: A Parallel Genetic Algorithm for Creating Virtual Portraits of Historical Figures// TASK Quarterly. -Vol. 16., iss. 1-2 (2012), s.145-162
- Sources of funding:
-
- COST_FREE
- Verified by:
- Gdańsk University of Technology
seen 63 times