红外多光谱遥感数据的校准  

Correction of Infrared Multispectral Remote Sensing Data

作　　者：方正[1] 林智乐[1] 钟硕 胡伟峰 陈思媛[1] FANG Zheng;LIN Zhi-le;ZHONG Shuo;HU Wei-feng;CHEN Si-yuan(Department of Instrumentation and Electrical Engineering,Xiamen University,Xiamen 361102,China)

机构地区：[1]厦门大学仪器与电气系,福建厦门361102

出　　处：《红外》2018年第10期40-43,48,共5页Infrared

摘　　要：主要介绍了一种对由特定红外图谱一体化系统采集的光谱数据进行校准与分析的方法。在实验部分,首先对得到的设定已知特定温度的黑体红外光谱数据进行归类和整理,然后在最大保留特征信号的条件下利用不同的滤波算法与滤波窗宽尽量消除信号噪声,并采用偏最小二乘算法对所得信号进行校准,使之与理论值相符。最后利用校准结果构建数学模型,并考虑主成分个数这一概念,检验模型准确度。通过数据分析可知,由这个特定红外图谱一体化系统采集的光谱数据经偏最小二乘算法校正后与理论值相近,此时能够拟合出校正数列与波数之间的函数关系,且受温度影响较小。A correction and analysis method for the spectral data collected by a specific infrared spectrum integrated system is presented. In the experiment, the infrared spectral data of a blackbody at a known specific temperature are firstly classified and sorted. The signal noises are eliminated as much as possible with different filtering algorithms and filter window widths on condition that the characteristic signals are maximally reserved. Then, the signals obtained are corrected by a partial least square algorithm so as to let them agree with the theoretical value. Finally, the correction results are used to construct a mathematical model and the concept of confidence is taken into account to verify the accuracy of the model. It is found in data analysis that the spectral data collected by the integrated infrared spectrum integrated system are close to the theoretical value after being corrected by the partial least-squares algorithm and the function relationship between the corrected series and the wave number can be fitted and is less affected by temperature.

关 键 词：红外光谱 校正 偏最小二乘 

分 类 号：O433[机械工程—光学工程]