Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4508563 | Engineering in Agriculture, Environment and Food | 2011 | 6 Pages |
Abstract
Ground-based hyperspectral remote sensing was applied to rice plants a week before harvesting in order to estimate the protein content of brown rice. When the accuracy was predicted through a full cross-validation, a 2007 model was estimated with R2 = 0.626, RMSEP = 0.311%, REP = 3.95% and a 2008 model was estimated with R2 = 0.819, RMSEP = 0.188%, and REP = 2.68%. When one model was used for prediction using the other year's data, the prediction error increased twice or more because the tendencies of the regression coefficients were different between the 2007 and 2008 models. When model prediction accuracy using two years' data was calculated by full cross-validation, R2, RMSEP, and REP were 0.850, 0.247% and 3.32%, respectively.
Related Topics
Life Sciences
Agricultural and Biological Sciences
Agronomy and Crop Science
Authors
Hiroyuki Onoyama, Chanseok Ryu, Masahiko Suguri, Michihisa Iida,