Research on Unstructured Road Detection Algorithm Based on the Machine Vision

This paper presents an unstructured road detection algorithm which can improve the detecting speed and accuracy of unstructured road detection. In this algorithm, first, we process the original images with the median value filter, and suppress the stochastic noise; Then an Otsu multi-threshold algorithm based on two-peak method for rapid image segmentation is used, which cause the division effect and the division time to be optimum; Finally the primary edge detection with the Canny operator and mathematics morphology are used, which can obtain the complete and clear path edge image, eventually enhance the accuracy of the path examination obviously. The simulation results show that the unstructured path examination method has the good characteristics of detecting speed and accuracy.