FTIR Spectroscopy for Seed Specie (Brachiaria Decumbens vs Brachiaria Ruziziensis) Classification

Abstract

The identification of Brachiaria seeds is typically performed by trained personnel through visual inspection, which can lead to misclassification and impact commercial operations. This study explores Fourier-transform infrared (FTIR) spectroscopy as an automatic method for classifying Brachiaria decumbens and Brachiaria ruziziensis seeds. Commercial seeds were cleaned and ground before analysis. The seed powder was characterized using an Agilent Cary 630 FTIR spectrometer with an ATR accessory, covering the 4000-650 cm⁻¹ range. FTIR data was preprocessed using the Modified Standard Normal Variate (MSNV) method, and seed classification was performed through Principal Component Analysis (PCA), which revealed clear clustering and separation. These results suggest that this methodology provides a robust tool for agricultural analysis.