The present invention provides a method for smoothly linking cameras to track ships, which realizes the detection and tracking of ship targets in continuous video images by using twin neural networks, obtains the ship target detection and tracking results relative to the image position, controls the camera pan and
tilt up and down, left and right, and the
focal length of the lens through the pan and tilt control protocol (PELCO‑D), and during the movement of the ship, the pan and tilt and lens control maintain smooth movement, the picture will not have obvious
jitter, and the ship is always in the center of the video picture. The present invention can be used in multiple water-related fields, such as the field of ship
traffic management in the maritime field, the supervision of fisheries in the
fishery field, and the monitoring of ships in the maritime police field; compared with the linked tracking of the ship position based on the vessel
traffic management system (VTS), the present invention uses image target detection and tracking, and the
image tracking effect is highly real-time and has a better tracking effect. It can be applied to scenes in the ship traffic
service system where the ship position
signal is updated slowly and the linked tracking effect is required to be high.