基于动态模糊积分的决策层融合识别算法

模糊积分是一种常用的信息融合方法,融合中最关键的问题是确定反映各信源重要程度的模糊测度。在此将该算法用于多传感器目标识别系统,首先介绍了Choquet模糊积分以及模糊测度的定义,再建立了基于动态模糊积分的决策层融合目标识别模型,将该过程转化为多个传感器的身份识别结果关于各自重要程度的广义Lebesgue积分。目前已有的确定模糊测度的方法几乎都只利用了训练样本的先验知识,适应性较差,难以全面地反映问题。该文在此基础上提出了一种基于动态模糊积分的决策层融合算法,可在判决过程中对结果进行动态的自适应修正,并给出了具体衡量各传感器重要程度的标准和方法。

Decision-making level fusion target recognition algorithm based on dynamic fuzzy integral

Fuzzy integral is a common information fusion algorithm. The key problem in the fusion algorithm is to determine the fuzzy measure reflecting the importance degree of each information source. In this paper,the algorithm is used in the multi-sensor target recognition system. The definitions of Choquet fuzzy integral and fuzzy measure are described. The target recogni-tion model of decision-making level fusion based on dynamic fuzzy integral is established. The process is converted to the each sensor recognition results related to the generalized Lebesgue integral on its importance degree. Almost all the current calculating methods to determine the fuzzy measure only utilize the prior knowledge of the training samples. Their adaptability is not good enough to reflect the existing problems roundly. A decision-making level fusion algorithm based on dynamic fuzzy integral is pro-posed in this paper,which can make dynamic self-adaption correction for the results in the judging process. The specific stan-dard and method of judging the importance degree of each sensor are offered in this paper.