文章基本信息

标题：A ROBUST FRAMEWORK TO DETECT MOVING VEHICLES IN DIFFERENT ROAD CONDITIONS IN INDIA
本地全文：下载
作者：Dr.S.APPAVU ALIAS BALAMURUGAN ; BALAJI GANESH RAJAGOPAL ; Dr.KUMAR PARASURAM 等
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2019
卷号：97
期号：1
页码：1-14
出版社：Journal of Theoretical and Applied
摘要：Traffic situation in India is a quite complex in nature when compared to the traffic models in other nations. It is very essential to model the traffic nature in Indian roadways, both rural and urban roads. Indian road conditions are predominantly occupies different classes of roads viz. single, double, multi-way, cross junctions etc. This research article addresses the different nature of Indian roads with an insight to model the traffic situations in different weather conditions also. The proposed system tries to solve the problem of counting and classifying the vehicles in Indian road conditions. The system uses color image based foreground moving object detection by preserving the color and model of the moving vehicles. The color image based background subtraction technique is supported by cascaded linear regression. The system also uses HoG for contour creation and extraction followed by morphological dilation to connect the missing pixels in the vehicle object. The framework uses adaptive Support Vector Machines to train and model the different classes of vehicles. It has been found that the proposed framework shows an accuracy of 92% in varying levels of traffic density, Illumination conditions.
关键词：Vehicle detection; Vehicle counting; Low quality video; Color image based background model; MoG; HoG; SVM Classifier