摘要
针对空间目标多光谱测温问题,提出了新的多光谱特征分析方法.该方法把目标红外辐射分为长、中、短三个波段,分析提取三个波段内的辐照度,以及长波段、中波段辐照度与短波段辐照度的比值,估算出目标温度.该方法避免了对目标光谱发射率的估计,有效减少了误差.最后基于系统参数,计算出有效辐射面积,并利用BP神经网络对四类目标进行识别试验,取得了较好的识别效果.
Based on the real multispectral data of space targets, a new way of feature- extraction is presented. That is, we separate the infrared radiation into three bands： long, middle, short. Then some features are analyzed as follows： the irradiance of three bands, irradiance＇s ratio of long, middle to short band, targets＇ temperature and radiant area. Finally a recognition test of four type targets with BP neural network demonstrates the good recognition resuit.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2005年第9期1564-1567,共4页
Systems Engineering and Electronics
关键词
红外多光谱
真温
空间目标识别
BP神经网络
IR multispectral
true temperature
recognition of space targets
BP neural network
作者简介
朱海峰（1979-）,男,硕士研究生,主要研究方向为空间红外目标识别.E-mail：kunta0089@sina.com
