A New Detection Framework for Stopped Vehicles in Urban Traffic Scenarios Using Robust Visual Features

In the current era of intelligence-based transportation systems, urban traffic analysis and congestion avoidance based upon movement of vehicles is obtaining huge attention because of its importance as well as complicated intrinsic components. However, in the present literature, congestion analysis on urban-traffic roads is primarily conducted based upon on-road vehicles, which could considerably change depending on the movement status of the vehicle. From another perspective, one of the primary factors for heavily congested traffic is that the presence of stopped vehicles (due to failure or other reasons, which occupy reasonable portion of road and also disturb the movement of nearby vehicles. In this manner, the stopped vehicles could affect traffic movement as well as congestion ratios to a larger extent. Due to these reasons, the detection of stopped vehicles followed by their tracking in urban traffic environments are very important to facilitate effective traffic flow designs. Based on these aspects, a novel framework for detecting the stopped vehicles on urban traffic scenes using robust SURF features is presented in this paper. The experimentations conducted on training, testing datasets clearly prove the reasonably good performance of the proposed framework, so that it could be successfully incorporated in congestion avoidance as well as evaluation systems.