多传感器目标识别的优化融合

对于工作在复杂环境下的多传感器目标识别系统,确保其稳健性和准确性的关键是有效处理被融合信息的不确定性。根据影响信息不确定性的因素,文章把传感器本地识别信息的可信度分为了统计可信度和环境可信度;采用最小二乘法和神经网络实现统计可信度的估计,自适应神经模糊推理实现环境可信度估计;并利用这两种可信度实现以一致理论为基础的多传感器目标识别的优化融合。经实验仿真证明,该融合方案是有效的。

Opitimed Fusion in Multi-sensor Target Recognition

For a multi-sensor target recognition system in the complex interference environment,it is important for steadiness and veracity of the system to deal with the uncertainty in fused information. According to the factors that affect the uncertainty, the reliability of local recognition was divided into statistical reliability and environmental reliability. The statistical reliability was estimated by neural network and least square method,and the environmental reliability was estimated by adaptive neural networks-fuzzy reasoning. Moreover, using the reliability, the paper presented the opitimed fusion based on consensus theory for multi-sensor target recognition. The simulation shows that the scheme is effective.