Article Preview
Top1. Introduction
Any Intelligent Transportation Systems (ITS) is dependent on reliable and real-time traffic details. Traffic Congestion (TraCo) information becomes the most essential and primary requirement for any ITS service application. Irregular events such as accidents, faulty traffic signals, breakdowns etc., cause unintended delay and congested scenario for commuters. Traffic congestion has also been a major problem affecting human daily lives, stabilizing economic and social developments (Reddy, 2018; Document on Transportation, 2020). It also tends to increase the pollution emission, fuel consumption, travel time, driver stress etc., (Kong 2016, Amudapuram, 2012). These situation causes higher real-time and reliable demand of TraCo information for ITS applications that monitor road traffic. Additionally, TraCo information becomes essential to applications that evaluate the performance or decide appropriate mitigation measures to aid comfortable commutations (Reddy, 2018). The congestion information can be enumerated from traffic status of a road segment that is described using vehicle traffic density, volume, speed, occupancy etc., (Duc-Binh 2018; Kong, 2016; ke, 2018; R.Wang, 2019; Wan-Xiang, 2018; Todd Litman, 2019). There are two major challenges in estimating traffic congestion. The first challenge is to facilitate the estimation over a large scale of road network. Several researches are attempting to address this issue and are divided into two categories (Kong, 2016; Duc-Binh, 2018). They include estimations of traffic status using infrastructure based technologies and Vehicular Adhoc Network (VANET) technology. Many congestion estimation and prediction algorithms have been adopted based on the data formalisation using either of the above two technology. The former approach include data acquisition methods using infrastructures such as the Global Positioning System (GPS), (Duc binh, 2018, Kong, 2016), sensor systems such as inductive loop detectors, infrared detectors (Bhaskar, 2015; Al-Naima) and traffic video archives (Reddy, 2018; Ke, 2018; Amudapuram, 2012). And later approach utilises Vehicle to Vehicle (V2V) communications from smart vehicles to gather traffic data (Duc-Binh, 2018; R.Wang, 2019; Badeddine, 2019; Manipriya, 2020). Infrastructure-based technologies have been the most easily available procurement process among the two technologies. Although VANET based methods have gained popularity in this research area and incorporated in various application, it is not widely available and has feasibility issues for heterogeneous traffic scenario. Also it has network related limitations such as redundancy, bandwidth and reliability issues, coverage, delay and inaccurate estimates which lead to inadequate precision and failure prone congestion estimations (Duc-Binh, 2018).
The second challenge in TraCo estimation is to ensure the quality of congestion information that depends on the accuracy, reliability and consistency of acquired data. Although infrastructure and VANET-based solutions have equal challenges in collecting and transferring data safely, sensor integration, overhead issue during rise in number of vehicles, etc., it is important to provide a simple and efficient solution to handle the problem to obtain precise congestion information. The existing systems do not address or resolve these shortcomings efficiently or systematically. Among many infrastructures available for traffic status enumeration, the use of the current traffic monitoring cameras serves several objectives without additional cost of installation, maintenance or equipment (Impedovo, 2019). Even though the data acquisition from traffic videos has been explored by researches over the past three decade, there are still challenges such as appropriate feature extraction, methods for detection and classification, video and camera specifications and properties etc., that need to be accomplished during vehicle detection (Zi Yang, 2018; Hanif, 2018; Boukerche, 2017;T. Bouwmans, 2019). This paper proposes a Traffic Congestion (TraCo) estimation process from traffic surveillance cameras using novel data acquisition from Multi-layer Continuous Virtual Loop (MCVL) framework or model. The proposed MCVL provides traffic status of the road segment focused by the surveillance camera using computer vision/ image processing techniques. The enumerated traffic statuses from MCVL are adapted in a fuzzy logic based method to address non deterministic problem of TraCo determination. The proposed system attempts to contribute successfully to enumerate traffic status required for ITS application with robust detection process for heterogeneous traffic condition and camera properties. The proposed work shows that the precision of the congestion information depends on the data acquired and has improved accuracy using MCVL.