摘要
在分析了以前的多传感器空间数据配准算法的特点和不足之后,提出了一种新的算法———基于聚类的数据配准:在多目标的情况下,先采用模糊c-均值法对传感器同一单帧量测数据进行聚类,得到的聚类中心作为各目标点的理想位置参数,再将由各目标点计算出的某一传感器误差值进行平均得到此传感器的误差估计,然后将各帧得到的误差估计再进行平均实现传感器配准。这种算法优点是实时性较强,与配准模型无关。最后给出了的仿真结果与分析。
A new multisensor registration algorithm via fuzzy c-mean cluster is presented after analysing some previous multisensor data registrations. In the case of multitarget, first, the same frame measurement data from multisensor are clustered via fuzzy c-mean algorithm to get the cluster center as the ideal position of targets. Second, certain sensor errors from these ideal position of targets are computed and then the errors mean are get to estimate the certain sensor error for this frame. Thirdly, these errors estimation from every frame mean are get to accomplish the multisensor registration.The advantages of this algorithm are better real-time and nothing with the registration model. In the end, the result and analyse of simulation is presented.