Abstract
The parallel computing algorithms are explored to improve the efficiency of image recognition with large database. The novel parallel version of the directed enumeration method (DEM) is proposed. The experimental study results in face recognition problem with FERET and Essex datasets are presented. We compare the performance of our parallel DEM with the original DEM and parallel implementations of the nearest neighbor rule and conventional Best Bin First (BBF) k-d tree. It is shown that the proposed method is characterized by increased computing efficiency (2-10 times in comparison with exhaustive search and the BBF) and lower error rate than the original DEM. © 2013 Springer-Verlag.
Author supplied keywords
Cite
CITATION STYLE
Savchenko, A. V. (2013). Real-time image recognition with the parallel directed enumeration method. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7963 LNCS, pp. 123–132). https://doi.org/10.1007/978-3-642-39402-7_13
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
