hyperspectral imaging system to determine volatile oils content of medicinal plants

This study was conducted to non-destructively and rapidly determine total volatile oils and therefore the quality of nepeta crispa willd samples using hyperspectral imaging system. nepeta crispa willd is a medicinal plant with high nutritional and therapeutical properties. the plant samples were procured from different geographical origins and processed using different drying methods (sun, shade, microwave, oven, vacuum oven, and freeze dryer). reflectance spectra (400-1000 nm) of the samples were collected and filtered using savitzky golay (sg) smoothing method. principal component analysis (pca) was used for a visual description of the samples and to select more effective wavelengths in order to do data reduction. two artificial intelligence models (radial basis function and multilayer perceptron neural network models) were created based on the total and more effective wavelengths (727-900nm) for volatile oils content prediction. results indicated that the hyperspectral imaging system coupled with artificial intelligence models was capable to predict the total volatile oils content of the nepeta crispa plant with a high degree of accuracy.