The water coastline detection approaches based on USV vision

Inorder to avoide the obstacle objects when the unmaned watersurface vehicle (USV) navigating on the water surface, a coastline detected approach is presented in this paper. By using the video camera on robot boat, the riparian under the surface of the natural environment is detected so as for autonomous obstacle avoidance and navigation. The algorithm is based on the traditional LOG edge detection method and improving this algorithm by stretching gray scale of the second derivative. Then convolution and threshold by Otsu algorithm are used to segement the coastline. The simulation results show that this approach can denoise the influence of the surface disturbances and complex background affects. It is faster and more accurate than the traditional LOG algorithm, and it can also improve the riparian positioning accuracy. Through the real applying this algorithm can efficiently and real time work in USV system.