Spectral Prior Knowledge and Its Use in the Remote Sensing Based Inversion of Vegetation Structure
光谱先验知识在植被结构遥感反演中的应用

Multiangular remote sensing observation can provide more spatial structure information that make it possible to retrieve the 3 D structural parameters of vegetation. Currently, most of the inversion algorithms only use one single band in the field of multiangular remote sensing, the correlation between bands hasn't been used efficiently for green vegetation. From the reflectance spectrum characteristic of green vegetation, we found that the basic shape of the spectrum can be used as prior knowledge in the inversion of vegetation structure. As a framework in this paper, spectral parameters were regenerated in the band of red, green and near infrared, their difference and ratio are used as the spectral prior knowledge in inversion. Through detailed simulation and inversion, it is clear from the statistical analysis that this method is more stable to observe noise than the single band based method which uses the averaged structural parameters inverted using one single band each time. It is also found that the inversion results tend to be closer to the ground truth than the single band based method after employing this kind of spectral prior knowledge.