Study On Methods Of Visual Monitoring And Illegal Sailing Detection Of Ships In Controlled Waterway

Controlled waterways are important river course in the upper reaches of theYangtze. Because of the limitation of the adverse geological and hydrologicalconditions, ships in a controlled waterway must receive commands from a shore signalstation and pass a waterway in order. If ships disobey the rules of navigation, and areprone to navigate in illegal behavior, which may cause a collision or stranding andaffect the normal operation of the social and economic life. At present, most of signalstations of controlled waterway give commands by VHF phones and monitor ships byhuman’s eyes, which can’t monitor ships effectively and record the direct evidencewhen ships sail illegally. In response to the development strategy of “digital waterway”and promoting the development of comprehensive transportation system of the YangtzeRiver, it is necessary to develop intelligent signal station with new technologies basedon the traditional monitoring methods in controlled waterways.For the requirements of visual and automated monitoring in the development ofintelligent signal stations and with the background of Shenbeizui controlled waterway inLuzhou of Sichuan province, this thesis tries to explore a visual method to monitor shipsand detect illegal sailing in controlled waterway. The major contents in this thesis canbe listed as follows:1. The thesis firstly analyzes geographic characteristics, weather feathers and noisesources from controlled waterways, especially noises from reflected shadows andtail-wave of ships, then studies the scene visibility assessment method, builds the visualships monitoring model in the controlled waterway and put forward our solutionframework.2. The thesis divides sailing trajectory detection into two problems: moving shipsdetection and tracking. For moving ships detection, the thesis focus on the method toweaken the noises from reflected shadows and tail-wave of ships and then proposes atwo detection method to filter the actual ships object; For ships tracking, this thesis firstresearch on the choice of better feathers including the color, gray and moment invariantthen choose moment invariants as the tracking feather of ships combined withprediction and tracking of Kalman filter to get positions of ships in a controlledwaterway.4. For illegal sailing detection of ships, the thesis introduces four forms of illegal sailing based on the navigation rules used in controlled waterway signal stations. Thenthe strategy of the illegal sailing detection is presented, based on the permitted sailingdirection determined by AIS data and the trajectory of ships within one or three cameraviews. At last, the primary testing results are given.The thesis tests the methods with video samples captured in the actual controlledwaterway. The result shows that the ships monitoring and the illegal sailing detectinghas been realized preliminarily.