Kendali Mutu Daun Jati Belanda (Guazuma ulmifolia) Menggunakan Pengolahan Citra dan Teknik Pengenalan Pola secara Kemometrik

Abstract

Quality of medicinal plants affects the quality of the herbs. The quality of medicinal plant, as jati belanda leaf (Guazuma ulmifolia) can be determined using image processing. The objectives of the research were to make spectral imaging as a method for quality control of jati belanda leaves by evaluating the quality based on various harvesting and combined with chemometric pattern recognition techniques (PCA and PLSDA). Jati belanda leaf image with various qualities were reconstructed as reflection spectrum using Wiener estimation. Goodness of spectrum reconstruction was presented from root-mean-square-error value of 36.65 % and goodness-of-fit coefficient value of 0.9576. The preprocessing technique was applied to the reflection spectra to get the best classification and prediction model. PCA analysis on the jati belanda leaves with data preprocessing using two first principal component (PC) was 100% able of the total variants in the data (PC1 92%, PC2 6%, PC3 1%, PC4 1%). PLSDA analysis produced 3 models. The models were sufficiently successful in differentiating the quality of jati belanda leaves.