A comparative study of pre-processing and classification methods for weed and crop discrimination
Résumé
In the context of precision agriculture, the problem of accurate weed and crop discrimination still remains a major issue due to uncontrolled lighting conditions and leaf orientations which produce differences in the measured reflectance spectrum. In this study, we tackle the problem of weed discrimination in wheat crop using hyperspectral images. To deal with spectrum differences, several preprocessing methods are evaluated. The influence of preprocessing is assessed with both linear PLS-LDA and non-linear SVMg (Gaussian Kernel) supervised classification methods.