Abstract
According to the Kyoto protocol, forest and non-forest area are distinguished using crown cover ratio. So the establishment of the method to estimate the crown cover ratio of forest is urgently necessary.
The formula which could be estimated crown cover ratio at in the high precision was made by using the spectral band ratio. In order to confirm the robustness of the formula, the next process produced 4 types stimulation image data. The pixels of the canopy crown were extracted from the ordinal CAST image first. Then, The pixels of the background except for the canopy crown were replaced with the spectral of water, soil, moss and the grass. The estimated canopy cover ratio had changed by the difference in the background types. From this result, the limit of the estimated formula became clear.
Then, the mixture classification technique was applied to estimate canopy cover ratio. When it was compared with the ground truth data, it showed a tendency to estimate it low as that result but the strong correlation existed between the truth and the estimated ratio. This research shows that the possibility to estimate the crown cover ratio accurately with mixture classification technique.
