V2I Implementation Using Artificial Vision and a Bayesian Classifier

Abstract

A procedure for classifying the state of vehicular traffic is implemented, using artificial vision tools, a Haar filter and a Bayesian classifier. The software developed simulates the operation of magnetic sensors distributed in the passable way. Digital image processing techniques allow vehicular detection in areas of complex traffic. The use of the Haar filter allows quantifying the number of vehicles parked and in circulation. The relevant information obtained in the video camera frames allows to establish a characteristic traffic vector. Subsequently, using a Bayesian classifier, the vector data is merged into four categories: zero traffic, low traffic, medium traffic and congested traffic. Finally, the results of the predictions are transmitted wirelessly between generic devices that simulate the known experience with the name of vehicle to infrastructure communication or V2I.