Comparative Study of PCA and LDA for Rice Seeds Quality Inspection

Contamination of rice seeds affects the crop quality, yield and price. Inspection of rice seeds for purity is a very important step for quality assessment. Promising results have been achieved using hyperspectral imaging (HSI) for classification of rice seeds. However, the relatively high number of spectral features in HSI data continues to pose problems during classification which necessitates the use of techniques like Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) for dimensionality reduction and feature extraction. This paper presents a comparative study of LDA and PCA as dimensionality reduction techniques for classification of rice seeds using hyperspectral imaging. The results of LDA and PCA on spectral features extracted from hyperspectral images were used for classification using a Random Forest (RF) classifier. Classification results shows that LDA is a superior dimensionality reduction technique to PCA for quality inspection of rice seeds using hyperspectral imaging.