Research On Navigation Technology Of Inland River Ship Base On Computer Vision

China is one of the largest inland river shipping countries in the world.Inland river shipping is an important part of China’s modern comprehensive system of transport.It has the advantages of large transportation capacity,low cost,little pollution,small land occupation and low energy consumption.In 2016,Yang Chuan Tang,Minister of Transport,proposed the overall goal of integrated transportation development at the National Transportation Working Conference:speeding up the construction of comprehensive traffic,smart traffic,green traffic and safe transportation,speeding up the construction of an integrated transportation infrastructure network.And by 2020,a safe and convenient,smooth and efficient,green and intelligent modern integrated transport system will be basically completed.The water transport industry has shown a new trend of development.With the development of inland river shipping,the number of ships has increased sharply and the traffic density has been increasing gradually.In addition,river navigation environment is complex and changeable.These factors seriously affect the safety of inland river shipping.However,most inland ships are ships with smaller weight.Based on cost considerations,they don’t install radar and other powerful navigation devices.Besides,the DWT also failed to meet the requirement of the IMO of AIS mandatory installation standards.Such ships navigating in river have big security risk.In order to improve navigation ability of inland small vessels,in this paper,a method of auxiliary navigation for inland river ships based on camera was proposed.And the related technology is studied.The main work is as follows:In order to improve navigation ability of inland small vessels,in this paper,a method of auxiliary navigation for inland river ships based on camera is proposed.And the related technology is studied.The main work is as follows:(1)For the video object in the ship,firstly,the image preprocessing is carried out by means of image grayscale,mean filter and histogram equalization,and an improved multi-directional edge detection method based on Sobel operator and rewitt operator is proposed to reduce the computational complexity of graphics and improve the image processing performance.(2)By setting a plurality of ROI(Region of Intrest)regions,the elimination of the region that blocks the detection of the target in the video image and the determination of the target region were realized.The Hog feature extraction method was used to reduce the interference of the water wave to the image,and the elimination of the water wave was realized.(3)According to the pinhole imaging principle and the camera’s projection relationship,the monocular geometric distance measurement method was used to locate the target vessel and the navigation function of computer vision was realized.