基于模糊特征属性参数最优融合的目标识别

Target Recognition Based on Optimal Fusion of Fuzzy Attributes

为了实现在不确定环境下的自动目标识别,提高系统的性能和可靠性。首先采用模糊数来刻画传感器的输出报告,将每个传感器报告用三角模糊数来表示;然后提出基于模糊特征属性参数的最优融合算法来实现信息融合,并将其应用到多传感器目标自动识别系统。融合算法以模糊信息相似性测度为基础,最优融合准则是:融合后的数据与各个传感器输入数据冲突应该最小。通过最优准则确定了各个传感器的权重以及融合结果,且融合结果与初始传感器权重向量和传感器报告融合的次序无关。文中给出了具体的算法流程和一个应用实例。

In order to realize automatic target recognition in uncertain environment, an optimal fusion of fuzzy attributes algorithm is presented to aggregate multi -source information. First, each sensor report is modeled as triangular fuzzy numbers. A simple similarity measure is introduced to determine the consensus degree of each sensor＇ report. Then, a criterion for optimal fusion is applied to combine all fuzzy information. The criterion is that the sum of weighted dissimilarity between the final fusion result and each sensor report should be minimized. By the iterative algorithm, the weight of each sensor and final fusion result can be determined. The optimal fusion algorithm takes advantage of many desired properties such as the order independence to the initial weight vector and the sensor report. Finally, the algorithm is illustrated numerically in target recognition problem.