Abstract
Multisensor Data fusion uses observations from networked multiple sensors achieves the same common air picture which has wide surveillance area. In ideal environment, a fundamental assumption is that registration errors are zero. However, in real environment, the registration errors are not zero. The plural targets are seen in the common air picture, even if the all sensors observe the same one target. In this paper, we propose new registration error estimation method which is composed of two registration error estimator. One is a coarse registration error estimatior which is based on corresponding point-to-point calibration for moving target, the other is a fine registration error estimatior which is based on kalman filter. As a simulation result, we confirmed that the proposed method achieves 82 point reduction of registration error estimation error than conventional method.
