Vehicle Tracking Using Morphological Properties for Traffic Modelling

Abstract

In the proposed method, an innovative image processing technique for vehicle tracking at a roundabout is described. Background subtraction is applied to get the objects (vehicles) in the foreground. The objects are thus obtained and tracked using morphological operations and object properties. In the video stream, tracking is established by tracing the center of the target object. In present case,four different directions of incoming traffic are considered and four vehicle classes are defined. Implementation of above mentioned method achieved promising result of accuracy greater than 90 % for moderate traffic conditions where occlusion is not an issue.