Author/Authors :
Abdul Rahim, Herlina Universiti Teknologi Malaysia - Protom-i Research Group, Infocomm Research Alliance, Control and Mechatronic Engineering Department, Malaysia , Seng, Chia Kim Universiti Teknologi Malaysia - Protom-i Research Group, Infocomm Research Alliance, Control and Mechatronic Engineering Department, Malaysia , Abdul Rahim, Ruzairi Universiti Teknologi Malaysia - Protom-i Research Group, Infocomm Research Alliance, Control and Mechatronic Engineering Department, Malaysia
Abstract :
This research investigates the use of predictive models and a low cost spectroscopy in non-invasive Soluble Solids Content (SSC) assessment. The challenge is to model complex and high-dimensional spectral data. Results indicate that the SSC prediction of pineapples harvested on different days using the proposed approach is promising. The second study evaluates common pre-processing practices. Findings indicate that the best accuracy would be achieved when visible spectrum was excluded, second order Savitzky-Golay derivative with the optimal filter length was used, and the absorbance transformation was avoided.
Keywords :
VIS , SWNIR , soluble solid contents , pineapples , principal component regression
