Vehicle Detection Using a Multi-Agent Vision-Based System

In this paper we propose a multi-agents system for vehicle detection in image. The goal of this system is to be able to localize vehicles in a given image. Developed agents are capable of detecting pre-specified shapes from processing this image. Cooperation involves communicating hypotheses and resolving conflicts between the interpretations of individual agents. Specifically in the proposed system, eight process agents, consisting of edge, contour, wheel, LPL (License-Plate Line), LPR (License-Plate Rectangle), PCV (Plate-Candidates Verification), and vehicle symmetry agents, were developed for vehicle detection in various outdoor scenes. In the testing data, there are 500 car blobs and 100 non-car blobs. We show through experiments that our system is 90.16% effective on Detecting vehicles in various outdoor scenes.