多源信息空间配准中的UT变换采样策略研究

Research on Sampling Strategies of UT Transformation for Space Alignment of Distributed Multi-Sensor Information

空间配准是分布式多传感器信息融合的基础。UT(UnscentedTransformation)变换以其非线性变换精度高,无需求导等优点得到广泛关注。目前,存在多种形式UT变换采样策略,而对这些采样策略缺乏比较分析。以多传感器量测信息的空间配准为应用,比较分析了对称性采样,最小偏度单形采样,以及超球体单形采样等多种UT变换采样策略。仿真结果表明,UT变换较一阶线性化近似方法更适于多传感器量测的空间配准,其中,对称性采样策略又优于其它采样策略。

Space alignment is the base of distributed multi-sensor information fusion. UT （Unscented Transformation） has been paid much attention because it is the second-order approximation at least without the requirement of derivable condition in nonlinear transformation. Now, multiple sampling strategies for the UT method has been developed while their comparison and analysis is absent. The comparison is proposed for the symmetric sampling, simplex min skew sampling and simplex sphere sampling strategies in the case of multi-sensor information space alignment from the sphere to Cartesian coordinates. The simulation shows that UT method based on any above strategies is superior to the first-order linearization approximation method while the symmetric sampling strategy is better than the others.