主观Bayes与神经网络组结合的信息融合目标识别

INFORMATION FUSION FOR TARGET RECOGNITION BASED ON SUBJECTIVE BAYES AND GROUP OF NEURAL NETWORK

针对神经网络在信息融合中的应用所存在的一些问题,受人的视神经网络的启发,提出了一种基于Bayes方法的神经网络组方法,并把它应用到了特征级融合中进行模式识别。仿真试验结果表明,和用单一的神经网络相比,利用神经网络组方法提高了网络收敛速度、缩短了网络的训练时间,提高了对目标的识别率,并且增强了对目标识别的稳定性。

This paper, enlightened by the optic nerve of human, presents a method--Group of Neural Network （GNN）--to solve the problems when the Neural Network is applied to information fusion, and applies it to the characteristic fusion to recognize the target. Simulations results show, compared with the Neural Network, this new method can improve the result of the recognition, and the stability of the recognition is enhanced, too.