A Fusion Detection Algorithm of Motional Ship in Bridge Collision Avoidance System

This paper proposed a detection algorithm for motional ships in complex inland ship channel based on opencv library. A ROI (Region Of Interest) for the input video frame is first selected. For the image within this ROI, we implemented edge detection and image segmentation. With these preprocessed images, Gaussian Mixture background subtraction method and three framess difference method are fused to detect the object in the foreground. Furthermore, the contour, the area of external rectangle, aspect ratio and other geometric features for the object are calculated. When these features of the object coincide with those of a ship, a rectangular box is added on the target. The experimental results showed that the algorithm can identify the motional a ship reliably, and assist the ship to sail safely.