基于联合相容分支定界的关联算法研究

Study on association algorithm based on joint compatibility branch and bound

联合相容分支定界算法(Joint Compatibility Branch and Bound,JCBB)充分考虑传感器量测之间的相关性和重新匹配关联的可能,但计算量随观测数目成指数增长。为优化其计算复杂度和关联准确度,以最近邻算法(Nearest Neighbour,NN)进行关联,对符合重复度和经过设定步数的情况使用JCBB进行特征匹配,并以互斥准则和最优准则来提高关联准确度。引入机器学习领域的评价测度对改进后算法和JCBB算法进行比较,结果表明,改进后的关联算法能够保证更好的关联准确度。

Joint Compatibility Branch and Bound Algorithm（JCBB） takes full account of the relevance and possible re-match association between sensor measurements,but the amount of computation increases exponentially with the number of observations. To optimize its computational complexity and association accuracy, we use the Nearest Neighbor algorithm（NN） to associate, when meet the repeatability and get to the number of steps set, employ JCBB for feature matching, and use exclusive criteria and optimal criteria to improve the relevance accuracy. we introduce evaluation measures in the field of machine learning to compare the improved algorithm and JCBB algorithm. The results show that the improved association algorithm ensures better association accuracy.