Texture analysis for road detection

Automatic road detection for the guiding of autonomous vehicles is a current problem. But the majority of studies are connected with the detection of yellow (or white) lines or road signs. The present authors attempt to remove these constraints by using the information of texture. The image to be processed is divided into elementary squares of 10*10 pixels. For each defined region, a vector which is composed of Galloway's parameters is calculated. The vector calculated makes possible the calculation of the resemblence between the region and the model of the road. The model is derived by the vector of Galloway's parameters for the elementary squares of the road images. The reliability and the possible extension of the method are discussed.<
>