基于光谱角度匹配的多光谱图像目标识别研究

Research on Multispectral Imaging Target Recognition Based on Spectral Angle Matching

使用地面推扫的多光谱成像设备,获取了三种典型情形下目标区域的多光谱图像。通过辐射校正、几何校正和光谱特征增强等预处理后,采用光谱角度匹配技术可成功将图像中与已知光谱角度一致的目标识别出来。结果表明,利用先验的光谱角度信息可以实现对多光谱图像中未知目标的分类、识别和提取。

The multispectral images of the target area in three typical scenarios are obtained using multispectral imaging equp-ment of ground sweeping.The targets can be identified successfully with the same spectral angel as known targets afer processed with radiation correction,geometric correction and spectral feature enhancement.The results indicate that it can be achieved classifi-cation,recognition and extreition of unknown targets multispectral images using prior spectral angle information.