Ray Tracingを用いた樹木モデルとパラメータ推定

抄録

This paper proposes a new model to estimate trees parameters, like tree size or density. Estimation of trees parameters using remote sensing data is required, for example, to evaluate greenhouse gas absorption or to manage forest. For this purpose, Geometric-Optical model has been proposed. This model calculates the geometry of a satellite and trees analytically, but it is difficult to consider trees over a slope area with oblique observation. Ray tracing technique is used in our new model to calculate the geometry accurately then the new model has much flexibility. Our new model also considers the spectral features of trees and grasses from a spectral library. Our model can be useful for spectral analysis. For the modeling, observation condition, environment condition, and observation target are defined. These include the satellite position, the sun position, target trees and grass. Then ray tracing is used to calculate the observed image geometrically. At that time, the model calculates the sunshade and shadow of trees also. After ray tracing, each pixel is assigned the spectral features from spectral library. The model calculates the brightness depending on the spectral band. Finally the pixel size is reduced to match the real satellite observation.
Trees parameters can be estimated to resolve an inverse problem based on this model using surface reconstruction method. Tree density and the size has been calculated from the simulated satellite image to evaluate the inversion.
The model was also evaluated by BOREAS data to estimate the tree distance by the surface reconstruction method. The estimated results was compared to the GIS data, and reached to the good results.