The Data Space Of The Multi-sensor Registration Algorithm And Engineering

This topic XXX comes from XXX institute project. The project's requirements in a multi-sensor data fusion target before the data preprocessing, data preprocessing mentioned here refers to a generic model of multi-sensor data fusion processing in the first level of data registration. This paper studies the multi-sensor data fusion technology, space for registration to get the engineering application of the registration model and algorithm.At present, most of the research results at home and abroad are a two-dimensional data within the registration. Registration of existing models and algorithms for the systematic errors are fixed. When the target farther away from the sensor, the existing problems with the registration model can not accurately reflected the real environmental model. When the system error in the change, commonly used parameter estimation algorithm can not be easily implemented in the computer.The actual battlefield environment to be able to build an accurate spatial registration model and solution algorithm, this paper mainly studies the following:1. To specify space for multi-sensor alignment task. Based on the actual environment, analyze the factors that affect the registration model is given a space with quasi-major source of systematic errors.2. For the systematic errors in the coordinate transformation in the process of changing circumstances, give a specific analysis. Analysis results as follows: Coordinate transformation impact on the system error is no law to follow. Mainly due to non-linear operation, the solution: the establishment of the target location in advance coordinate transformation error database, depending on the target location error of library search to compensate.3. For existing model algorithm was simulated and analyzed. The outcome: By least squares estimation, generalized least squares method and maximum likelihood estimation algorithms, such as comparison and found that random noise in the error distribution is known under the premise of maximum likelihood estimation algorithm is better than the previous two algorithms When the random error there is no a priori condition of circumstances, can only use least-squares parameter estimation algorithm. Adopting the above parameter estimation algorithm, system error fixed, because the project's needs systematic error is not constant, so the commonly used estimation algorithm does not apply.4.According to the distance and the establishment of a space with the registration model and Kalman filter algorithm is thinking of the systematic error parameters as a motion equation of state, in accordance with sensor coordinate transformations and space observation data with the task of quasi-systematic error parameters derived observation equation, use the Kalman filtering algorithm. Through scene modeling and simulation show that the model and algorithm. Concludes with the latest operational model gives a data link and radar model of the space between the alignment.5. Engineering. As the real-time intensity and a larger amount of data can only be implemented on microprocessors, choose the Power PC processor, in the VxWorks embedded operating system environment, the use of C language designed and implemented data link and radar with the space between the registration model.