文章基本信息

标题：Implementation of Optical Flow, Sliding Window and SVM for Vehicle Detection and Tracking
本地全文：下载
作者：Mohammad Baji ; Dr. I. SantiPrabha
期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2017
卷号：6
期号：9
页码：18452
DOI：10.15680/IJIRSET.2017.0609139
出版社：S&S Publications
摘要：Vehicle detection and tracking are two most challenging tasks of traffic surveillance in transport system.Traditional vehicle detection and tacking methods are computationally expensive and become inefficient in case wherelight intensity is low and occlusion of vehicles is high. This paper includes detection of moving vehicles, counting ofvehicles and tracking of detected vehicles. Here detection is carried out by Optical Flow Method in which a movingvehicle is extracted from the relative motion by segmenting the region representing the same optical flows.The opticalflow image uses Horn-Schunk technique in finding the optimum parameters for providing the smoothness. The detectedvehicles are tracked using Linear Support Vector Machine Classifier (Linear SVM) with a set of features likeHistograms of Oriented Gradients (HOG), Histograms of color features and Geometric features of vehicles, withthisdetails the classifier is trained.This trained classifier is used to search for vehicles in a video, based on the sliding –window technique then its heat map is generated to know the position of a vehicle. Combine this heat maps gives theexact location of the vehicle in images. Experimental results demonstrate that the proposed method is effective indetection&counting and tracking of vehicles.
关键词：Detection and Tracking; Horn-Schunk method; Optical Flow; Morphological operation; HOG; SVM;Sliding-Window technique.