Automated classification based on video data at intersections with heavy pedestrian and bicycle traffic: Methodology and application

Pedestrians and cyclists are amongst the most vulnerable road users. Pedestrian and cyclist collisions involving motor-vehicles result in high injury and fatality rates for these two modes. Data for pedestrian and cyclist activity at intersections such as volumes, speeds, and space–time trajectories are essential in the field of transportation in general, and road safety in particular. However, automated data collection for these two road user types remains a challenge. Due to the constant change of orientation and appearance of pedestrians and cyclists, detecting and tracking them using video sensors is a difficult task. This is perhaps one of the main reasons why automated data collection methods are more advanced for motorized traffic. This paper presents a method based on Histogram of Oriented Gradients to extract features of an image box containing the tracked object and Support Vector Machine to classify moving objects in crowded traffic scenes. Moving objects are classified into three categories: pedestrians, cyclists, and motor vehicles. The proposed methodology is composed of three steps: (i) detecting and tracking each moving object in video data, (ii) classifying each object according to its appearance in each frame, and (iii) computing the probability of belonging to each class based on both object appearance and speed. For the last step, Bayes’ rule is used to fuse appearance and speed in order to predict the object class. Using video datasets collected in different intersections, the methodology was built and tested. The developed methodology achieved an overall classification accuracy of greater than 88%. However, the classification accuracy varies across modes and is highest for vehicles and lower for pedestrians and cyclists. The applicability of the proposed methodology is illustrated using a simple case study to analyze cyclist–vehicle conflicts at intersections with and without bicycle facilities.     

A traffic accident morphology diagnostic model based on a rough set decision tree

ABSTRACT

To build a traffic safety feature model and to quantify accident influences caused by some traffic violation behaviors of drivers, an accident diagnostic decision-making model is established. For the purpose of diagnosing accident morphologies, rough set theory is applied and the influence of traffic factors of different accident morphologies is quantified through calculating the degree of attribute importance, selecting core traffic factors and adopting a C4.5 decision tree algorithm. In the paper, road traffic accident data from 2008 to 2013 in Anhui Province are used. Typical rules are selected, targeted strategy proposals are put forward, and then, a scientific and reasonable diagnostic basis is provided for the diagnosis of traffic safety risks and the prediction of potential traffic accidents.     

The authors in this issue

This paper considers the main road-traffic parameters that determine air pollution, i.e. the total volume of traffic, road speeds and the composition of the vehicle fleet. Changes in the amounts of pollutants emitted, and the importance of each of the three parameters, have been computed by using a traffic assignment model, which also represents emission factors. The types of policies that may be implemented to reduce the environmental impact of transport are then considered. The study demonstrates, for example, that the impact of a deterioration in traffic conditions is limited in comparison with the effect of forecast increases in traffic and improvements in the environmental performance of vehicles. As a consequence, if cities and urban transport are to achieve sustainable development, urban expansion must take place in a controlled way.     

A model for the visualization and analysis of elevator traffic

ABSTRACT

Analysis of elevator traffic in high rise buildings is critical to the performance evaluation of elevator group control systems (EGCS). Elevator dispatching methods or parking algorithms in an EGCS can be designed or modified according to analyses of traffic flow. However, interpretation of traffic flow based solely on numerical data may not be explicit and transparent for EGCS experts as well as for other non-expert building administration. In this study, we present a model for visualization and analysis of elevator traffic. First, we present an alternative approach for traffic analysis which we call route visualization. In the proposed approach, we initially decompose elevator traffic into its component parts and investigate each component independently. Then, using superposition of components we obtain a reconstructed model of overall traffic. This modeling approach provides component-based traffic analysis and representation of routes with intensities through data visualization. In the second part we introduce a multi-dimensional analysis of time parameters in ECGS. This approach provides a comparative analysis of several control algorithms such as dispatch or park algorithms for different combinations of traffic components.     

Pedestrian detection for traffic safety based on Accumulate Binary Haar features and improved deep belief network algorithm

ABSTRACT

In order to improve traffic safety and protect pedestrians, an improved and efficient pedestrian detection method for auto driver assistance systems is proposed. Firstly, an improved Accumulate Binary Haar (ABH) feature extraction algorithm is proposed. In this novel feature, Haar features keep only the ordinal relationship named by binary Haar features. Then, the feature brings in the idea of a Local Binary Pattern (LBP), assembling several neighboring binary Haar features to improve discriminating power and reduce the effect of illumination. Next, a pedestrian classification method based on an improved deep belief network (DBN) classification algorithm is proposed. An improved method of input is constructed using a Restricted Bolzmann Machine (RBM) with T distribution function visible layer nodes, which can convert information on pedestrian features to a Bernoulli distribution, and the Bernoulli distribution can then be used for recognition. In addition, a middle layer of the RBM structure is created, which achieves data transfer between the hidden layer structure and keeps the key information. Finally, the cost-sensitive Support Vector Machine (SVM) classifier is used for the output of the classifier, which could address the class-imbalance problem. Extensive experiments show that the improved DBN pedestrian detection method is better than other shallow classic algorithms, and the proposed method is effective and sufficiently feasible for pedestrian detection in complex urban environments.     

A Multiple Criteria Decision‐Making Approach for Prioritizing Accident Hotspots in the Absence of Crash Data

Abstract

In an efficient transportation system, traffic safety is an important issue and it is influenced by many factors. In a country like Iran, until now safety improvements are mainly concentrated on road engineering activities, without much attention for vehicle technology or driving behaviour. One important aspect of road safety engineering activities is the so‐called treatment of hotspots or dangerous accident locations. Until recently, accident hotspots were identified and remedied by the esxperts’ personal judgements and a handful of statistics without taking into account other important factors such as geometric and traffic conditions of the road network. This paper therefore aims to define and identify the criteria for accident hotspots, then giving a value to each criterion in order to develop a model to prioritize accident hotspots when traffic accident data is not available. To do this, the ‘Delphi’ method has been adopted and a prioritization model is produced by the use of a ‘Multiple Criteria Decision‐Making’ method. The procedure is illustrated on a collection of 20 road sections in Iran. In addition, the model is validated against an existing database of road sections containing safe locations and hotspots. Finally, a sensitivity analysis is carried out on the proposed method.     

Arterial travel time estimation based on vehicle re-identification using wireless magnetic sensors

A practical system is described for the real-time estimation of travel time across an arterial segment with multiple intersections. The system relies on matching vehicle signatures from wireless sensors. The sensors provide a noisy magnetic signature of a vehicle and the precise time when it crosses the sensors. A match (re-identification) of signatures at two locations gives the corresponding travel time of the vehicle. The travel times for all matched vehicles yield the travel time distribution. Matching results can be processed to provide other important arterial performance measures including capacity, volume/capacity ratio, queue lengths, and number of vehicles in the link. The matching algorithm is based on a statistical model of the signatures. The statistical model itself is estimated from the data, and does not require measurement of ‘ground truth’. The procedure does not require measurements of signal settings; in fact, signal settings can be inferred from the matched vehicle results. The procedure is tested on a 1.5 km (0.9 mile)-long segment of San Pablo Avenue in Albany, CA, under different traffic conditions. The segment is divided into three links: one link spans four intersections, and two links each span one intersection.     

Recognition and tracking of spatial–temporal congested traffic patterns on freeways

The two models FOTO (Forecasting of Traffic Objects) and ASDA (Automatische Staudynamikanalyse: Automatic Tracking of Moving Traffic Jams) for the automatic recognition and tracking of congested spatial–temporal traffic flow patterns on freeways are presented. The models are based on a spatial–temporal traffic phase classification made in the three-phase traffic theory by Kerner. In this traffic theory, in congested traffic two different phases are distinguished: “wide moving jam” and “synchronized flow”. The model FOTO is devoted to the identification of traffic phases and to the tracking of synchronized flow. The model ASDA is devoted to the tracking of the propagation of moving jams. The general approach and the different extensions of the models FOTO and ASDA are explained in detail. It is stressed that the models FOTO and ASDA perform without any validation of model parameters in different environmental and traffic conditions. Results of the online application of the models FOTO and ASDA at the TCC (Traffic Control Center) of Hessen near Frankfurt (Germany) are presented and evaluated.     

Vehicle detection grammars with partial occlusion handling for traffic surveillance

Traffic surveillance is an important topic in intelligent transportation systems (ITS). Robust vehicle detection is one challenging problem for complex traffic surveillance. In this paper, we propose an efficient vehicle detection method by designing vehicle detection grammars and handling partial occlusion. The grammar model is implemented by novel detection grammars, including structure, deformation and pairwise SVM grammars. First, the vehicle is divided into its constitute parts, called semantic parts, which can represent the vehicle effectively. To increase the robustness of part detection, the semantic parts are represented by their detection score maps. The semantic parts are further divided into sub-parts automatically. The two-layer division of the vehicle is modeled into a grammar model. Then, the grammar model is trained by a designed training procedure to get ideal grammar parameters, including appearance models and grammar productions. After that, vehicle detection is executed by a designed detection procedure with respect to the grammar model. Finally, the issue of vehicle occlusion is handled by designing and training specific grammars. The strategy adopted by our method is first to divide the vehicle into the semantic parts and sub-parts, then to train the grammar productions for semantic parts and sub-parts by introducing novel pairwise SVM grammars and finally to detect the vehicle by applying the trained grammars. Experiments in practical urban scenarios are carried out for complex traffic surveillance. It can be shown that our method adapts to partial occlusion and various challenging cases.     

Comparing prediction methods for maritime accidents

ABSTRACT

The purpose of maritime accident prediction is to reasonably forecast an accident occurring in the future. In determining the level of maritime traffic management safety, it is important to analyze development trends of existing traffic conditions. Common prediction methods for maritime accidents include regression analysis, grey system models (GM) and exponential smoothing. In this study, a brief introduction is provided that discusses the aforementioned prediction models, including the associated methods and characteristics of each analysis, which form the basis for an attempt to apply a residual error correction model designed to optimize the grey system model. Based on the results, in which the model is verified using two different types of maritime accident data (linear smooth type and random-fluctuation type, respectively), the prediction accuracy and the applicability were validated. A discussion is then presented on how to apply the Markov model as a way to optimize the grey system model. This method, which proved to be correct in terms of prediction accuracy and applicability, is explored through empirical analysis. Although the accuracy of the residual error correction model is usually higher than the accuracy of the original GM (1,1), the effect of the Markov correction model is not always superior to the original GM (1,1). In addition, the accuracy of the former model depends on the characteristics of the original data, the status partition and the determination method for the status transition matrix.     

Fine-grained OD estimation with automated zoning and sparsity regularisation

Given a road network, a fundamental object of interest is the matrix of origin destination (OD) flows. Estimation of this matrix involves at least three sub-problems: (i) determining a suitable set of traffic analysis zones, (ii) the formulation of an optimisation problem to determine the OD matrix, and (iii) a means of evaluating a candidate estimate of the OD matrix. This paper describes a means of addressing each of these concerns. We propose to automatically uncover a suitable set of traffic analysis zones based on observed link flows. We then employ regularisation to encourage the estimation of a sparse OD matrix. We finally propose to evaluate a candidate OD matrix based on its predictive power on held out link flows. Analysis of our approach on a real-world transport network reveals that it discovers automated zones that accurately capture regions of interest in the network, and a corresponding OD matrix that accurately predicts observed link flows.     

Network models in traffic management and control

Abstract

The context for network modelling in traffic management and control is described in terms of the current area‐wide nature of traffic management and the range of objectives to which it can contribute. Representation of a road system and traffic management measures in terms of nodes and links and parameters associated with them is described. It is shown that the pattern of traffic has to be represented not only in terms of flows on links of the network but also in terms of numbers of movements per unit time between points of entry to and points of exit from the area being modelled. In modelling so far, these numbers of movements are regarded as given, but the routes taken are estimated by traffic assignment. Models can so far be used for comparison of a range of given schemes and for optimization of traffic control within a scheme. Variation over time is a central feature of the modelling, and this requires the use of time‐dependent queueing theory, and the specification of numbers of movements for a succession of periods of between 10 and 30 minutes. Theoretical approaches to the resulting problems of modelling and optimization are discussed, and the ways in which these are supplemented by heuristic methods in currently available models is described. Some requirements‐for further research are outlined.     

The Benefits for Rail Freight Customers of Vertical Separation and Open Access

Abstract

This paper reviews and analyses the benefits for rail freight customers of the two principal models for introducing competition in main line railway networks: (1) the vertical separation of infrastructure from operations; and (2) the introduction of competition providing other operators with open access to the network. The paper shows that traffic growth has generally been higher in those European countries that have liberalized most. It then examines in detail the impact of restructuring in Great Britain in terms of the competition, freight rates and traffic growth. The paper concludes that vertical separation benefits freight customers more than just open access.     

A novel variable selection method based on frequent pattern tree for real-time traffic accident risk prediction

With the availability of large volumes of real-time traffic flow data along with traffic accident information, there is a renewed interest in the development of models for the real-time prediction of traffic accident risk. One challenge, however, is that the available data are usually complex, noisy, and even misleading. This raises the question of how to select the most important explanatory variables to achieve an acceptable level of accuracy for real-time traffic accident risk prediction. To address this, the present paper proposes a novel Frequent Pattern tree (FP tree) based variable selection method. The method works by first identifying all the frequent patterns in the traffic accident dataset. Next, for each frequent pattern, we introduce a new metric, herein referred to as the Relative Object Purity Ratio (ROPR). The ROPR is then used to calculate the importance score of each explanatory variable which in turn can be used for ranking and selecting the variables that contribute most to explaining the accident patterns. To demonstrate the advantages of the proposed variable selection method, the study develops two traffic accident risk prediction models, based on accident data collected on interstate highway I-64 in Virginia, namely a k-nearest neighbor model and a Bayesian network. Prior to model development, two variable selection methods are utilized: (1) the FP tree based method proposed in this paper; and (2) the random forest method, a widely used variable selection method, which is used as the base case for comparison. The results show that the FP tree based accident risk prediction models perform better than the random forest based models, regardless of the type of prediction models (i.e. k-nearest neighbor or Bayesian network), the settings of their parameters, and the types of datasets used for model training and testing. The best model found is a FP tree based Bayesian network model that can predict 61.11% of accidents while having a false alarm rate of 38.16%. These results compare very favorably with other accident prediction models reported in the literature.     

A GIS‐based traffic analysis zone design: technique

The main purpose of this paper is to develop an efficient method to design traffic analysis zones (TAZs), which is necessary for implementing a planning process with Geographic Information System (GIS) for Transportation (GIS‐T), using statistical spatial data analyses and GIS technology. The major roles of GIS in this method are: (1) to produce basic spatial units (BSUs) with topological data structure; (2) to integrate various procedures during the TAZ generation including computer program routines; and (3) to visualize the output of each TAZ generation. One of the most significant reasons for obtaining well‐defined TAZs is the fact that they are defined at the outset of transportation demand modeling, used from trip generation to trip assignment, and will ultimately affect transportation policy decisions.

Toward obtaining well‐defined TAZs, this paper concentrates on two important constraints: homogeneity and contiguity. Iterative partitioning technique is adopted to promote the optimum homogeneity of generated TAZs, while a contiguity checking algorithm is developed to ensure contiguous TAZs are generated by the iterative partitioning technique.     

A systems approach to area traffic control

Since the first pilot scheme for area‐traffic control was introduced in the city of Montreal (1959–60), computer control of traffic in urban areas through the adaptation of existing traffic‐signal systems has been provided to an increasing extent. This area of work may pose problems for the professional traffic engineer whose background in computer technology and general digital electronics may be limited.

In considering the engineering implications of such schemes a systems approach is important and is adopted here. Three existing and representative schemes are briefly mentioned in order to outline basic features. A more detailed examination of the various system elements follows with mention of data collection and transmission, and the role of the control computer.

The paper continues with a reconsideration of the three representative schemes in the light of the detailed treatment of system components. It concludes with a tentative assessment of the present position of area traffic control schemes and some suggestions as to the future development.     

基于视频的行人检测算法研究

文章提出了一种基于视频图像的行人检测算法。通过动态阈值目标分割算法获得运动目标,同时将分割结果块化处理,以减少算法时间复杂度。使用最小外接矩形获得目标的高宽比,结合基于质心的目标跟踪方法获得目标物体运动速度,进而对行人事件进行检测。实验结果表明,该方法能准确地检测交通场景中的行人事件,有效地解决抖动、光线及积水等因素对行人目标检测的影响,具有良好的实时性和准确性。     

Statistical mechanics for analytic planning: an application to domestic air traffic in China

Abstract

Statistical mechanics has shown its usefulness when assessing the topology of many networks, including those of infrastructure. Its principles take into account the large-scale and network-wide effects of changes in its key parameters, which in turn may provide critical input when planning for infrastructure projects. One objective would be to modify the pattern of capacity expansions inside a system to make it less exposed to local shortfalls in demand. To illustrate our point, we shall use domestic air traffic in China; airports are spatially distributed and they also need to respond to the potential demand that they face locally. Airlines that control parts of the traffic system are identified as agents. A relationship between the agent's behavior and the system-wide level of variance in traffic flows can be established by regression analysis. It is shown how intervention on these agents would reduce negative traffic variance while enhancing a more balanced, less costly growth of the system itself.     

Hybrid artificial neural network and locally weighted regression models for lane-based short-term urban traffic flow forecasting

ABSTRACT

In recent years, there has been considerable research interest in short-term traffic flow forecasting. However, forecasting models offering a high accuracy at a fine temporal resolution (e.g. 1 or 5?min) and lane level are still rare. In this study, a combination of genetic algorithm, neural network and locally weighted regression is used to achieve optimal prediction under various input and traffic settings. The genetically optimized artificial neural network (GA-ANN) and locally weighted regression (GA-LWR) models are developed and tested, with the former forecasting traffic flow every 5-min within a 30-min period and the latter for forecasting traffic flow of a particular 5-min period of each for four lanes of an urban arterial road in Beijing, China. In particular, for morning peak and off-peak traffic flow prediction, the GA-ANN 5-min traffic flow model results in average errors of 3–5% and most 95th percentile errors of 7–14% for each of the four lanes; for the peak and off-peak time traffic flow predictions, the GA-LWR 5-min traffic flow model results in average errors of 2–4% and most 95th percentile errors are lower than 10% for each of the four lanes. When compared to previous models that usually offer average errors greater than 6–15%, such empirical findings should be of interest to and instrumental for transportation authorities to incorporate in their city- or state-wide Advanced Traveller Information Systems (ATIS).     

Locating inspection facilities in traffic networks: an artificial intelligence approach

Abstract

In order for traffic authorities to attempt to prevent drink driving, check truck weight limits, driver hours and service regulations, hazardous leaks from trucks, and vehicle equipment safety, we need to find answers to the following questions: (a) What should be the total number of inspection stations in the traffic network? and (b) Where should these facilities be located? This paper develops a model to determine the locations of uncapacitated inspection stations in a traffic network. We analyze two different model formulations: a single-objective optimization problem and a multi-objective optimization problem. The problems are solved by the Bee Colony Optimization (BCO) method. The BCO algorithm belongs to the class of stochastic swarm optimization methods, inspired by the foraging habits of bees in the natural environment. The BCO algorithm is able to obtain the optimal value of objective functions in all test problems. The CPU times required to find the best solutions by the BCO are found to be acceptable.