Research On Inland River Shipping Tracking And Track Fusion Method

With the vigorous development in our country inland river shipping, water transport is becoming more and more arduous. As inland river ports throughput increased dramatically, ship traffic and shipping density also will continue to increase, which lead to a shipping accident risk is also increasing, it seems particularly important to dynamic tracking and real-time monitoring of inland river ships sailing. In recent years, vessel traffic management system (VTS) radar, automatic identification system (AIS), electronic chart display information system (ECDIS) and other intelligent monitoring equipment have been used in inland river is maritime supervision. The radar is the best choice in tracking the ship, but it will lose their tracking ability and can’t identify the ship type in the blind zone; AIS can provide the ship dynamic and static information, but also can identify properly type of ship, because of its high cost of installation, in Inland River300tons of the following small ships not equipped it; with the mature technology of video monitoring technology, and install them cheap, makes up for the blank in the inland river ship real-time monitoring, which provides the possibility of a comprehensive monitoring of inland river ship, but it can only obtain the information in the limited channel.Therefore, this thesis fusion the navigation ship trajectory of the above three methods to improve the ship tracking precision, which based on the ship target recognition and tracking method, it has important acacemic and application value. The main work and innovation of this thesis is as follows:1) The recognition method of inland waterway ships based on videoThe thesis proposed a kind of moving target recognition method based on improvement of the adjacent frame and Hu moment invariants.At first; the2D wavelet transform was performed for filtering the noise. Then the moving targets are extracted by the Canny operator and the adjacent frame difference. An improved method of invariant moment’s extraction is proposed to extract seven invariants characteristic value, and a BP neural network recognition algorithm based on optimizing the weights of additional momentum factor is designed. The sample of the torque characteristic value as input vector of neural network, and get accurate classification results, it is benefit the next target dynamic tracking.2) The tracking method of inland waterway ships based on videoThe thesis proposed an adaptive bandwidth mean-shift tracking algorithm and an update method of window width using scale detection. The window width can change follow the size of movement target, it solves the problem of fixed window width of the traditional mean-shift tracking algorithm, to achieve the goal of real-time tracking.3) The Multi sensor information fusion of video, AIS and radarAfter analyzing the ship information characteristics collected by AIS and radar, considering all possible result in the process of fusion two kinds of information, the thesis established a recognition framework of the ship location decision. It sets evidence credibility according to their importance using Kalman filter algorithm, gives a right credibility formula, determines the dynamic reliability of AIS and radar, and fuses all information by combination rules. Then revising the AIS and radar fusion results apply get the video of ship motion trajectory, to eliminate the abnormal data and get a more accurate ship motion trajectory and improve the moving target tracking accuracy.