Research On Detection Model Of Vessel Deck Location

Waterway takes a quite important role of modern transportation because of its unique advantages,especially on the transportation of long distance and heavy goods.However,in recent years,due to the increasing number of inland vessels,the waterway is becoming more and more busy,with the overloading and over drafting happened frequently,so that the safety pressure of waterway transportation has been rising continuously.At present,maritime departments determine whether the navigable vessel is overloaded or not by measuring its freeboard manually.Manual method is inefficient and costs much of time,so automatic measurement of vessel's freeboard has become an important topic to concern,in order to help maritime departments to protect inland rights.With the rapid development of modern information technology,the use of various high-precision sensors is becoming more and more popular,which provides powerful tools for vessel's freeboard measurement.This dissertation mainly do research on extracting vessel's deck from laser scanning data and calculating freeboard of inland vessels.Inevitably there will be various types of noise in laser scanning,in order to obtain accurate results of freeboard caculation,preprocess work on scanning data is essential.The scanning data contains a lot of information about the surface of the object,so this dissertation uses the three-dimensional reconstruction method to describe the surface of the vessels,in order to observe features of target.Water surface and vessel surface are the main components of laser data,so firstly establish the height distribution model of water surface to separate water and vessel.A sample of scanning data may contains multiple vessels,so use statistical filtering and projection to segment multiple targets,and keep the filtered and smoothed single vessel as result of preprocess.The distance between waterface and deck is the freeboard of vessels,so in addition to acquire water data,the deck data is also an important part in vessel's freeboard measurement.According to the geometrical information and position information of deck in the vessel,this dissertation proposes two kinds of deck extraction algorithm based on spatial feature.Line detection and plane detection are the key technology of target recognition in three-dimensional space.In this dissertation,the detection of vessel's deck is based on the straight line detection by hough transform,and palne detection by the neighborhood feature and the normal vector feature.After deck detection,the distance between waterface and deck the is the freeboard of vessel.This dissertation proposes a fusing algorithm which combines the two method based on both features of line and plane,and experiments show that the fusing algorithm achieves higher measurement accuracy and better real-time detection effect.