Traffic monitoring is a really broad concept and it has numerous usages and implementations. The goal of the system described in this paper is to monitor street traffic. The results can be widely used:
• makes the traffic load measurable on the monitored sections of the road
• can be used as a reference for further urban development
• gives immediate feedback on the effect of changing traffic management
• makes the estimation of air pollution caused by the traffic easier
The Internet of Things is the next important step in the history of the World Wide Web. The concept behind this is that billions of sensors, vehicles and other devices connected to the Internet communicate with each other exchanging a tremendous amount of data. A part of this vision is the Smart City which based on the Internet of Things makes the life and the city administration easier in the urban areas. This concept gives space to this traffic monitoring system because the devices will be placed in the case of street lights and will do the processing locally and transfer only the results to the server.
The system does the task of traffic monitoring by using a camera and real-time image processing. During the development frequent tests needed for the verification of the image processing algorithms. These tests are vital, because the algorithms are so complex that this is the easiest way to evaluate the efficiency of a modification.
A part of the tests are done on a pre-recorded set of videos, where we already know the correct results therefore the accuracy of the system is measurable. The other tests are done on the actual device where we can monitor the operation in real time. It is really important to make the devices easy to update with the latest version even there are multiple devices available because the mass-testing of the devices is an important goal.
For the proper operation the calibration of every single device is a must. This has to be done one by one, because the given parameters (camera distortion, distance from the ground, the angle of the camera, the position of the lanes, etc.) are different for each sensor and these are significantly affect the accuracy of the system.
The goal of this paper is the presentation of the already implemented solutions in terms of diagnostics, software update and testing and the demonstration of the test environment and high level diagnostics functions implemented by myself.