Fast Road Edge Detection Method Based on Canny Operator and Dynamic Thresholds

This paper proposes an improved road edge detection method based on the Canny operator and dynamic thresholds to improve the detection speed while better preserving the details of edges. First, the local Gaussian filtering is applied to the pixels on the road edges in the image, and the Otsu method is used to adaptively determines the local high and low thresholds to improve the edge connectivity. Second, the gradient calculation is performed with eight directional gradient templates of the Sobel operator to improve the edge localization effect. Third, the gradient calculation results of the points are compared with the thresholds to determine whether they are edge points. After the edge points are identified, the seed-growing method is used to find continuous road edges. Experiments are conducted and the results demonstrate that the proposed algorithm can improve the detection speed of road edges by 5 to 7 times compared with the traditional Canny operator. And by using the dynamic thresholds, the adaptability of the algorithm and the connectivity of road edges are improved.