To deal with big parking lots and buildingswhich are misclassified as road networks, a directional texturedetector is developed to distinguish different types of objectsaccording to their textures in different directions. Thedirectional texture detector measures the pixel grey valuevariance along the central lines in each of four directions of anoperation window. If all the variances in four directions aresmaller than a certain value, it can be concluded that the objectwithin this window is homogeneous. Therefore, this object canbe considered as a non-road object and can be removed. Figure7 shows the input image and the processed result. The classifiedroad network (Figure 7a) is the input. The result of the segmentfiltering is shown in Figure 7b. After segments filtering andedge-aided segmentation, the final road network is extracted(Figure 7c).
đang được dịch, vui lòng đợi..
