Sea-sky-line detection method for visible image based on semantic segmentation and Hough Transform
Sea-sky-line detection is a key technology for the visual perception of intelligent navigation of ships.To address the challenge of sea-sky line detection being interfered with by ship,waves,and weather in maritime traffic images,a detection method based on semantic segmentation and Hough Transform is proposed.The image is input into the Mo-SegNet image semantic segmentation algorithm improved based on SegNet network for training,and the image features are learned;the Canny detection algorithm is used to detect the contours in the semantically segmented image;and the Hough Transform is used to fit the edge pixels to get the sea-sky-line detection results.In Scene 1,its detection accuracy and intersection and concatenation ratio reach 99.28％ and 74.85％,respectively;in Scene 2,its detection IoU (Intersection over Union) reach 98.60％ and 59.41％,respectively;in Scene 3,its detection IoU reach 99.31％ and 77.19％,respectively,and the experimental results show that the proposed algorithm combining the semantic segmentation and the Hoff Transform has high accuracy and high precision in the sea-sky-line detection.with high accuracy and effectiveness.
NETL
NSTL
万方数据
