Road Surface Recognition System Based on Its Picture

A grip of the road finding seems to be very solid problem in case of self-driven cars. It can be done by recognition of surface the car is moving on. The photos of the road taken while driving a vehicle can be the solid data source for the detection process. It is rather difficult to extract information from pictures that provide data necessary for classifier to distinguish patterns. Apparently simple preprocessing is not enough since a neural network was not able to learn anything. To resolve the problem of the road recognition entirely new picture preprocessing type has been developed. It fits circles of uniformly brightness areas then counts them and measures their sizes. The learning of the multilayer perceptron realised by such data gives very good results. The new way of extracting data from pictures is a promising solution and was named as ”Growing Bubbles Algorithm”. The algorithm was implemented as part of a real system to support the on-line driver decision. The system was tested in the real car in real traffic with very promising results.