
OBJECT DETECTION USING AM-FM FEATURES
Author(s) -
Kanneboyigaraju,
B. Eswara Reddy,
P. Chandra Sekhar Reddy
Publication year - 2011
Publication title -
international journal of smart sensors and ad hoc networks
Language(s) - English
Resource type - Journals
ISSN - 2248-9738
DOI - 10.47893/ijssan.2011.1018
Subject(s) - artificial intelligence , computer science , pattern recognition (psychology) , computer vision , object (grammar) , feature (linguistics) , object detection , image (mathematics) , feature extraction , philosophy , linguistics
This paper presents a template-based approach to detect objects of interest from real images. We rely on AM-FM models and specifically, on the Dominant Component Analysis (DCA) for feature extraction. We incorporate the results from AM-FM models for object detection. In order to detect the object of interest from real images patches are introduced. In order to find the degree of match between the patch and template, the AM-FM features are calculated. To find the correlation between the template and image patch, mean and standard deviation of image patch and template are calculated. If this correlation value exceeds a preset detection threshold, we declare that patch containsthe object of interest. The combination of AM-FM features and template based object detection produces efficacious results.