Multi-sensor Data Fusion And Target Tracking Algorithm

The multisensors data fusion is a comprehensive discipline whose theories, techniques and methods are carried on processing to the multiple source information.Data fusion techniques can achieve a better performance of detecting, tracking, classifying targets and rise the system's reliability or survivability effectively by combining data from multiple sensors or multiple kinds of sensors.This paper concerns the problem of multi-target tracking and the application of neural network in targets tracking by data fusion technology. The main works are summarized as follows:1. First, we introduce the definition, fundamental principle, and methods of multisensor data fusion system. Then we discuss the multisensor data fusion function model and its application scopes.2. The basic principle and type of maneuvering target tracking is introduced in the dissertation. It lays the foundation for the behind content further discussion.3. In this chapter,we have a research on a asynchronous data fusion algorithm of distributed multisensor systems. A new fusion algorithm based on the new model and traditional Kalman filter is proposed.For each sensor,we can estimate and update every state by obtaining orderly measures in this period.Then the local state estimates of all sensors will be transported to the fusion center where the next state estimate may be got by global information. Using of computer simulation at last,we compare the results utilizing the new algorithm with those based on time calibrated method via estimate accuracy, the good performance arising from this new approach has been effectively validated.4. We study the application of neural network in targets tracking. On the basic of the discussion of flight path fusion based radar and infrared,we added the BP neural network,and simulated.The results of the simulation showed BP neural network can get high precision with smaller error.