Design And Implementation Of Multi-means Data Fusion System Based On Real-time Coputation And Offline Mining

The theoretical basis of multi-sensor data fusion concept has been more mature after years,and with the rapid development of information and computer technology,large-scale,various and high-precious sensors are used in more scenes,resulting in a higher requirements of data fusion technology.In the multi-sensor network architecture,data fusion system,as a processing layer,provides support for the upper application.When the diversity of sensor equipment and detection targets makes the collected information more massive,multi-source and heterogeneous,whether the data can be processed efficiently and accurately has become the key to ensure the accuracy of sensor network application.Although there is already a precedent for big data technology for data processing in data fusion,the major technology used in this area can only meet one requirements of one computing model:higher real-time requirements use real-time streaming computing model,while higher accuracy requirements use offline processing.A single computing model can not meet the changing needs of data processing,therefore,studying a dual-computing architecture is a necessary requirement to promote the development of multi-sensor data fusion technology.In view of the above problems,this paper,based on the background of the data fusion requirements in the field of maritime surveillance,investigates the two computing models widely used in big data technology,relies on multilevel storage media,designs and implements a system multi-means data fusion processing system of real-time computing and offline mining,where the real?time calculation part,based on the flow calculation framework,is responsible for associating short-term description information of different dimensions to the same target,and the offline mining part forms the long-term target tracks by integrating the results of real-time calculation,and excavates and analyzes the motion track,expecting to obtain the target's behavioral characteristic description to provide decision support for determining the target behavior in real time part.In order to verify the effectiveness of the system,the real scene as the test background to verify the functional requirements of the system,and test the performance of the system optimization.The experimental results show that the system designed in this paper can meet various needs in maritime surveillance,and has good performance in performance,which proves that the system proposed in this paper is feasible and versatile.