摘要
提出了一种基于虚拟多传感器融合技术的红外目标识别方法.文中利用傅里叶描述器提取目标形状的边缘特征以及辐射特性的六个特征量,采用多个人工神经网络对来自单一传感器的目标利用不同特征分别识别,再利用D-S证据推理将各个网络的识别结果进行决策级融合.仿真实验结果表明,该方法提高了识别率和识别结果的可靠性.
A pscudo multi-sensor fusion approach is proposed in this paper for target identification in infrared images. The Fourier descriptors of targets boundary as shape features and the six gray parameters as radiation features are extracted. A novel strategy is investigated which utilizes artificial neural networks (ANN) to recognize the same IR target from a single sensor respectively by means of different features, then fuses the information at decision level. The simulation results of experiment show that the approach can improve the recognition efficiency and reliability.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2002年第3期355-358,共4页
Pattern Recognition and Artificial Intelligence
基金
重点实验室开放基金
