The quality of public roads depends highly on the evaluation of their condition. A part of the evaluation is detecting the pavement distresses, measuring their parameters, and classificating them, based on the measurements. Because of the lack of IT support it is done manually, or with a technology with high expense, although it could be done with image processing faster, more cost-effective or even more accurately.
The purpose of this thesis is to develop an algorithm, which can detect, measure and classificate pavement distresses automatically, or with minimal user interaction. Along this it can offer a user interface, where the user can inspect, correct, and evaluate the results.
Firstly I introduce the cv4sensorhub framework and the Open Source Computer Vision Library (OpenCV). I show two ways of their recognition based on the color histogram of their images, and I evaluate the results. I present the requirements for taking samples and the influence of environmental effects on the results. I introduce how pavement distresses can be classified, how their parameters can be measured and classified based on the parameters. The final application contains the previous features and offers a graphical user interface to evaluate the results and make corrections if necessary.