基于改进霍夫变换的水面无人船水界线检测方法

准确的水界线检测是水面无人船(unmanned surfaced vessel,USV)导航和自主避障的关键;同时也为水面图像处理提供有效区域。针对复杂背景下水界线检测实时性和鲁棒性欠佳的现象,提出一种改进的水界线检测方法。首先,使用Canny边缘检测提取图像的边缘信息;然后,使用基于边缘梯度信息的Hough变换,限定参与水界线拟合的边缘像素点的梯度方向;并删除对水界线没有贡献的边缘像素点,同时缩小累加器空间;最后,使用加权长度确定水界线,以此抑制标准Hough变换对纹理性区域边缘像素点造成的虚假峰值。实验结果表明该算法可准确、高效地检测出水界线。

Waterline Detection Method for USV Based on Improved Hough Transform

Accurate waterline detection is the key technique of navigation and obstacle avoidance for unmanned surfaced vessel (USV), and it also provides an effective area for surface image processing. Aiming at the poor real-time and robustness of waterline detection in complex background, this paper proposes an improved waterline detection method. First, the Canny edge detection method is used to extract the edge information of the image; then, the Hough transform based on the edge gradient is used to limit the gradient directions of the edge pixels that involved in the waterline fitting, and the edge pixels that do not contribute to the waterline are deleted, and the accumulator space is reduced as also. Finally, the weighted length is used to determine the waterline, which effectively suppresses the false peak caused by the traditional Hough transform on the textural edge pixels. The experimental results show that the algorithm can detect the waterline accurately and efficiently.