Applying Petri Nets to Approximation of the Euclidean Distance with the Example of SIFT

Abstract. SIFT (Scale Invariant Feature Transform) is a complex image processing procedure for matching objects or patterns in images. Involving the computation of Euclidean distances in high dimensional space of many pairs of points and matching them against a threshold, this work proposes a speedup for the procedure utilizing colored Petri nets. Being sped up more pairs can be evaluated within reasonable time leading to better overall results.

• Type: Journal
• Year: 2015
• Where: APN
• Authors: Jan Henrik Röwekamp, Michael Haustermann
• Keywords: APN 2015, Artificial Intelligence