Comparative Analysis of Vehicle Detection Methods using OpenCV

Abstract / Excerpt:

An efficient way of detecting vehicles for a system in traffic management or road monitoring is crucial. If a system can't handle detection of vehicles effectively it can lead to many problems. Computer vision techniques allow you to detect and track vehicles in real time. You can use different techniques to detect vehicles in various conditions. This study will focus on comparing different object detection techniques which are: Double differencing method, appearance-based approach, feature-based approach and edge-based tracking and implement them to be able to identify a more effective vehicle detection method. This aims to provide an efficient way of detecting vehicles in different conditions for a better and more effective traffic management system, road monitoring system and other vehicle related systems. The results show that double differencing method has the highest accuracy rate out of the other 3 object detection techniques; it performs with an average accuracy rate of 78% and an average percent error rate of 22%. It performs well on day conditions including day + rain condition. Individual results for each object detection techniques are presented in this paper.

Source InstitutionAteneo de Davao University
UnitComputer Science
AuthorsRay Anthony Y. Saavedra
Page Count11
Place of PublicationDavao City
Original Publication DateOctober 1, 2013
Tags Analysis, Comparative, Detection, Methods, OpenCV, Vehicle