|
Aiming at the problem that precise recognition cannot be achieved in ground-based measurement of high orbit satellite targets(GEO), this study carried out the measurement and recognition of GEO targets by ground-based hyperspectral imaging system. By establishing a spatial target scattering spectral imaging model, the scattering imaging process of GEO targets is analyzed from the physical perspective. Based on this, the measurement experiments of GEO targets are carried out, and the spectral video images and spectral data of four groups of targets are obtained. Through the calibration and normalization of spectral data, the optical features of different targets are further analyzed combined with spectral image features, which provides technical support for target classification and recognition. At the same time, the fitting and inversion of target spectral data is carried out based on partial least squares fitting method. The relationship between the target spectrum and the target typical material spectrum is analyzed. Through fitting, it is found that the typical material categories in the four targets are CF, AsGa, Si and Golde. Among them, CF and Golden in target 1 and target 4 are relatively large, about 80% of the overall spectral content. The proportion of CF in Goals 2 and 3 is about 75%. Different proportions of typical materials are reflected, which to some extent explains the different composition structure of the target or the different instantaneous projection pose, and provides a basis for the recognition of target pose.
|
