HAAR Characteristics-Based Traffic Volume Method Measurement for Street Intersections

HAAR Characteristics-Based Traffic Volume Method Measurement for Street Intersections

Santiago Morales (Universidad Nacional de Colombia, Colombia), César Pedraza Bonilla (Universidad Nacional de Colombia, Colombia) and Felix Vega (Universidad Nacional de Colombia, Colombia)
Copyright: © 2020 |Pages: 28
DOI: 10.4018/978-1-7998-1839-7.ch011


Traffic volume is an important measurement to design mobility strategies in cities such as traffic light configuration, civil engineering works, and others. This variable can be determined through different manual and automatic strategies. However, some street intersections, such as traffic circles, are difficult to determine their traffic volume and origin-destination matrices. In the case of manual strategies, it is difficult to count every single car in a mid to large-size traffic circle. On the other hand, automatic strategies can be difficult to develop because it is necessary to detect, track, and count vehicles that change position inside an intersection. This chapter presents a vehicle counting method to determine traffic volume and origin-destination matrix for traffic circle intersections using two main algorithms, Viola-Jones for detection and on-line boosting for tracking. The method is validated with an implementation applied to a top view video of a large-size traffic circle. The video is processed manually, and a comparison is presented.
Chapter Preview


Below is the background of the problem from the point of view of the measurement and modeling of vehicular traffic in cities, and the tools currently used to collect primary information that allows designing such models. The second point of view illustrates the most common current technologies used to detect and track objects through image processing. Finally, the central problem of the article is presented, i.e. the difficulty of obtaining the origin-destination matrix (known as OD Matrix) at intersections.

Traffic Volume Measurement

As the volume of vehicles increases in large cities, it becomes necessary to measure, analyze, model and evaluate vehicular traffic, in order to more accurately determine their behavior on the roads of modern cities. This allows making wise decisions that contribute to the improvement of mobility. For example, traffic models influence the design of roads, vehicular intersections, or the optimization of traffic signal timing (de la Rocha, 2010).

The traffic problem began to be analyzed in the 50s, when mathematical models began to compare vehicular flow with the movement of particles in fluids (Hoogendoorn & Bovy, 2001). Since then, the subject has been thoroughly studied and debated. However, it is necessary to collect information in real environments to evaluate and contrast it with reality.

Manual counting is within the most commonly used methods, in which trained personnel are located on roads or intersections to count vehicles (Instituto Nacional de Vías (INVIAS), 2013). However, this method is expensive, is not performed regularly, and manual counts are error-prone. Other alternatives replace the human being with automatic devices such as pneumatic sensors, infrared sensors, or inductive loops, all installed on the road (Federal Highway Administration, 2013). These alternatives reduce human-induced error but are expensive and invasive with the infrastructure.

Complete Chapter List

Search this Book: