Analysis of SVM Technique for Vehicle Classification

Authors

  • Kavitha N, Dr.Chandrappa D N

Abstract

The analysis of vehicular traffic data plays a vital role in effective traffic management. Detection of moving vehicle in a video still remains challenging task due to complexity of the scene and illumination variations. In this paper we propose simplified approach to detect and classify the vehicles for a video input. Initially foreground is extracted from background frame using Gaussian Mixture Model (GMM), the extracted features such as corner detection, color transform and edge detection are compared with training phase to verify whether vehicle pixel is present or not. Finally Support Vector Machine (SVM) is used to classify the vehicles by the extracted features with training database and comparison of contours.

Keywords: SVM technique, Gaussian Mixture Model (GMM), foreground, color transform, traffic management

Published

2020-12-08

Issue

Section

Articles