The road safety became a global concern with the release of the “World Report on Road Traffic Injury Prevention” by the United Nations and the World Health Organization in April2004. Some important mistakes were detected like “insufficient attention to the design of traffic systems”. This statement confirms, validates and justifies research the reason to develop and design traffic monitoring systems to help reduce the actual traffic accidents rate.

In this situation, strategic traffic management is a key factor in nowadays. This report encouraged traffic authorities to coordinate traffic management and monitoring focusing in Road Safety problems. The Intelligent Transport Systems (ITS) present a solution in traffic management and user mobility. The development and use of different applications and technologies in ITS systems increase road safety by incorporating breakthrough solutions in different levels of management and control.

In this thesis it is exposed, discussed and developed an effective, efficient and multipurpose ITS solution that would support the traffic authorities, providing plentiful, diverse, accurate and in real time information about the road situation. This solution is designed to be compatible with any commercial sensor that uses the area detection principle, this is, laser scanners. 

The investigation developed in this thesis is based in ITS sensors and the traffic control center (TCC) needs. Therefore, it has been able to define the requirements and architecture of the intermediate block that will communicate both areas. This will help to implement and develop applications framed in the scope of eSafety such as: shadow tolls, travelers’ information systems, highways monitoring and management systems, access controls, incidents management …

Regarding the hardware, it was designed and implemented a multi-sensor platform, able to work with several commercial equipments and different information formats. The hardware unit carries out the signal pre-treatment, but only data captured by the sensor with information from vehicle detection is sent to the TCC. Effective signal-processing software analyzes the signal and treated undesired effects due to the nature of laser scanner, like, lateral views or lost reflections. 

The methodology used for vehicle classification is described by means of pattern recognition techniques. In the learning process certain features from the vehicle silhouette are extracted. The selection of the features of the vehicle silhouette used in the discrimination has been key for success in the classification process. In the learning process are extracted the characteristics of the group that allow the discrimination between classes on the basis of certain predictive parameters. The success of the classification process is based on a correct selection of these parameters. Since this is a case of supervised learning solved by using nonparametric modelling techniques, classification trees have been used. This classification or decision trees define, firstly, the basis of each one of the categories, and, secondly, are used to classify new samples.

Finally, by means of the Bootstrappping technique – adaptive resampling - the estimators have calculated the goodness of the classification system. This technique is particularly advantageous when used in conjunction with decision trees. This is because decision tree algorithms are relatively efficient for samples with many dimensions; and decision trees tend to have larger variance components than other methods.

As a result of sophisticated data analysis, a highly accurate vehicle detection and classification is achieved with extremely low false alarm rates. Tests presented good results: namely, detection rates of 97.89%, precision rates in detection of 99.694% and classification accuracy rates of 94.24%. The system provides a set of sensor, controller, software, and communications modules for detecting and classifying vehicles in real time, acquiring and storing their silhouette and 3D image; as well as other important traffic parameters (density, flow rate, occupancy or traffic volume) shown in a frontend at the TCC.