Uncertain data mining from spectra library under Bayesian network model

Uncertainty is an inherent property of Remotely Sensed data. Under the architecture of Bayesian network, which can integrate the quantitative and qualitative knowledge into a comprehensive probabilistic knowledge representation and inference environment, this paper presents a model for data mining from spectra library. Using the filed measured data to drive the model, we obtain the crop structure variables such as Leaf Area Index (LAI) information, e.g. its probability distribution, which can be looked as the priori knowledge of the parameter during the process of inversion.