Research On Multi-Sensor Data Fusion For Ship Target Detection

Maritime management plays an important role in national economy and security, and ship surveillance is one of main work in maritime management. Obtaining the accurate ship distribution image and estimating ship motion situation in a surveillance area are the foundation of ship management. Synthetic Aperture Radar (SAR), High Frequency Surface Wave Radar (HFSWR) and Automatic Identification System (AIS) are three main sensors for surveillance in large maritime areas. Space-borne SAR has wide coverage and high resolution, and does not have weather restrictions. However, it can detect ships only at the time of the satellite overpass. HFSWR has the advantages of continuous surveillance, scan long range and direct velocity estimation, and the disadvantages of low space resolution. AIS transmits ship information, i.e., position, ship length, ship width, velocity and heading, for ship collision avoidance, but not all ships carry AIS equipment. These three sensors have their merits and shortcomings in ship surveillance. An accurate ship traffic image cannot be obtained via one sensor alone. The main porpose of reasearch is how to fuse and use these sensors'measurement to improve the ship detection accracy. Ship point target association, target tracking and track-to-track association of Space-borne SAR, HFSWR and AIS measurement is investiagted in this dissertation. The main contents are as follows:1. In order to solve the fusion problem of HFSWR and AIS target point tracks in dense ship distribution, a point tracks association algorithm using Jonker-Volgenant-Castanon (JVC) global optimal matching is proposed. Firstly, the AIS target point tracks, which are measured in World Geodetic System 1984 (WGS84), are mapped into the polar coordinate system based on HFSWR station. Then, the measurement models of HFSWR and AIS are established, and the pair gating and iterative search algorithm are used to divide the whole measurment set into fesible association subset. Finally, the HFSWR and AIS point tracks of every fesible association subset are associated with the JVC global optimal association algorithm. The simulation experimental results indicate that the proposed algorithm is superior to the Nearest Neighbor (NN) algorithm and Joint Probability Data Association (JPDA) association algorithm in the association accuracy, and the association time is less than the JPDA algorithm. Moreover, the experimental results of real-life target point tracks in three years demonstrate that the proposed method is feasible.2. For the assocation of Space-borne SAR, HFSWR and AIS point track, a Maximum Likelihood (ML) association algorithm with multi-feature improvements is proposed to increase detection accuracy and reduce false alarms. The tested features are position, size, heading and velocity. Based on the ship measurement model, the problem of data association for SAR, HFSWR and AIS is formulated as a multi-dimensional assignment problem. In the data assignment process, JVC and Lagrangian relaxation algorithms are applied. Simulation results show that the algorithm proposed here can improve the association accuracy compared with the NN and the position-only ML algorithms, using the additional features of length and velocity. Real data experiments illustrate that the algorithm can enhance target identification and reduce false alarms.3. In order to overcome the deficiency that vessels missed in tracking with single frequency HFSWR, a vessel fusion tracking algorithm with dual-frequency HFSWR and calibrated by AIS is proposed. AIS information is used to estimate and correct the HFSWR system bias for each frequency. First, the point measurements of AIS from cooperative vessels are associated with the measurements of HFSWR using the JVC assignment algorithm. From the associated results of cooperative vessels, the system biases of dual-frequency HFSWR are estimated and corrected. Then, based on the corrected dual-frequency HFSWR data, vessels are tracked by dual-frequency fusion JPDA-Unscented Kalman Filter (UKF) algorithm. Experiment results using real-life detection data showed that the proposed method is efficient at tracking the vessels in real time and can improve the tracking capability and accuracy compared with tracking with single frequency data.4. In order to solve the problem of HFSWR and AIS track-to-Track association, a multi-model track-to-track association algorithm is proposed based on the track features. First, the AIS tracks are classified into two categories, which are the nearly straight tracks and change tracks, by the rotation rate and course over ground (COG).Then, change tracks are segmented into nearly straight tracks at the big turning point. Finally, the weighed similarity of radial range, azimuth and radial velocity between HFSWR and AIS tracks are selected as the association vector, and tracks are associated by global track-to-track assocation algorithm. Real data experiments illustrate that the proposed algorithm can improve the association accuracy in dense track association. For the change track, track segmention and association can effectively solve the segmented track correlation problem caused by tracking target lost in HFSWR tracking.