Grey Relation Estimating Pattern of Soil Organic Matter with Residual Modification Based on Hyper-spectral Data

As to the low prediction accuracy in estimating the content of soil organic matter with hyper-spectral technology, at first, based on the uncertainties of hyper-spectral information and the non-time data series, a new grey weighted relation degree model is put forward and the grey relation estimating model is established. Then, by making full use of recognition error, a modified model is given and the grey relation estimation pattern with the residual modification is proposed. At last, the pattern proposed in this paper is applied to the hyper-spectral estimation of soil organic matter that was collected from Tai'an City in Shandong Province. The results indicate that the testing samples' average relative error is 5.563%, while the average relative errors of the classic grey relation pattern and linear regression model are 10.128% and 10.567% respectively, the average relative error of forecasting samples is 6.994%. The application example shows that the pattern proposed in this paper is valid.