Incremental Multi-Feature Tensor Subspace Learning Based Smart Traffic Control System and Traffic Density Calculation Using Image Processing

Abstract

In times, urban centers are growing at a high rate. Growing with them is a road traffic jam. Traffic jams, especially at peak hours, became routine. As a result, traffic management is one of the foremost pressing issues in today's towns. Several alternatives are being sought to affect the matter. These include expanding road networks, regulating the number of vehicles on the roads, and deployment of Intelligent Transportation Systems (ITSs). Aside from the ITSs, the opposite alternatives (however significant) have many practical challenges in their implementation. ITSs have supported a good range of technologies like loop sensors and video surveillance systems. Vision-based ITSs have proved advantageous over the standard methods supported loop sensors. In these modern systems, video surveillance cameras are installed along the roads and road intersections where they're wont to collect traffic data. The info is then analyzed to get traffic parameters like road traffic density. This paper presents a comfortable and stylish approach for estimating the road traffic density during daytime using image processing and computer vision algorithms. The video data collected is first weakened into frames, which are then preprocessed during a series of steps. Finally, the vehicles are detected and extracted from the pictures and Density estimated. The traffic density is then obtained because of the number of vehicles per unit area of the road section. The proposed approach was implemented in MATLAB R2013a and average vehicle detection accuracy of 96.0% and 82.1% were achieved for fast-paced and slow-moving traffic scenes.