Vehicle classification based on structural and local features.
Suryanti Yunita Anggrelly.
Date of Issue2011
School of Electrical and Electronic Engineering
Institute for Infocomm Research, Singapore
Object classification research has been moving towards invariant features extraction and development of a robust framework for object modeling and recognition. However, only a few works have been reported in implementing them in a real-time traffic surveillance system, in particular for vehicle classification task. We propose a hierarchical method using structural and local features for vehicle classification in an automated real-time traffic surveillance system. In the first stage, major planes in the vehicle image are extracted to build the structural configuration of the vehicles. Descriptors obtained using Scale Invariant Feature Transform (SIFT) algorithm are used as the local features in the second stage of classification. Each class of vehicles is represented by a number of images selected using our proposed template selection method. Keypoints from these templates are further reduced to remove redundant keypoints. The proposed method was tested on images taken from a real-time traffic surveillance database and performed well on the vehicle classification.
DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Control engineering