Utilizing KNN for Estimating Lignin in Rice Bran through Color Imagery with PCA Preprocessing

Abstract

Feed is essential for enhancing livestock production, particularly in maintaining animal health and stamina. Rice bran is commonly used as animal feed; however, its quality can decline when mixed with other ingredients, such as rice husks. The addition of rice husks to rice bran increases the levels of crude fiber and lignin, which are difficult for livestock to digest and can lead to health issues. This mixing can be assessed by estimating the lignin content through the phloroglucinol dye reaction. This study aimed to estimate the lignin content in a mixture of rice bran and rice husks using the dye reaction and the resulting color images. The images were captured using the red-green-blue (RGB) color model. A feature extraction technique called principal component analysis (PCA) was employed on each RGB component. The results from the PCA were subsequently classified using the k-Nearest Neighbor (KNN) algorithm. The findings indicated that the red (R) color component yielded the highest classification accuracy of 77.27%.