多重分光画像における要素スペクトルの適応的推定
スポンサーリンク
概要
- 論文の詳細を見る
Spectral unmixing is a method which intends to estimate the proportion of each category in one pixel from multispectral data. The usual methods assume that the component spectra are known and are derived from training data. However, the spectra derived from the training data are not always correctly representing the spectral characteristics of the categories within the objective area. Here we propose a method of adaptive spectral unmixing which estimates suitable component spectra from the data themselves. By using the adaptively estimated spectra, we can estimate the proportion of components even if the spectral characteristics change with the objective area.
- 帝京平成大学の論文
- 2001-06-30