Title :
An architectural hardware scheme for the generation of LG graphs for real-time image analysis
Author :
Michalopoulos, Kostas ; Tsakalakis, Michail ; Bourbakis, Nikolaos
Author_Institution :
ATR Center, Wright State Univ., Dayton, OH, USA
Abstract :
This paper presents a hardware architecture scheme that implements the local-global (LG) graph methodology. The LG graph method has the capability to efficiently represent the context of a segmented image and hold its structure independent to rotation, translation and scaling. The LG graph method has been successfully used on object/pattern recognition, image understanding, face/facial expression recognition, biometrics representation and recognition.
Keywords :
graph theory; image segmentation; LG graph generation; biometrics recognition; biometrics representation; face expression recognition; facial expression recognition; hardware architecture scheme; image segmentation; image understanding; local-global graph methodology; object recognition; pattern recognition; real-time image analysis; Computer architecture; Corporate acquisitions; Hardware; Image segmentation; Merging; Real-time systems; Shape; Image processing/analysis; LG graphs; parallel processing;
Conference_Titel :
Digital Signal Processing (DSP), 2013 18th International Conference on
Conference_Location :
Fira
DOI :
10.1109/ICDSP.2013.6622727