Robust provenance classification of curcumae radix (Yujin) using nir-hsi: Preprocessing, nested CV and morphology effects

Key Points

• Provenance classification is enhanced through robust techniques involving NIR-HSI, leading to significant accuracy improvements.
• Nested cross-validation shows a remarkable 85% accuracy in classification, confirming method effectiveness.
• Morphological features are critical; variations influence classification, enhancing the understanding of heritable traits.
• This study implies that improved classification methods may lead to better quality control in herbal medicine applications.