S17 - Session O2 - Hyperspectral imaging for predicting the chemical composition of fresh-cut pineapple (Ananas comosus).

Authors: Manuela Zude-Sasse *, Kaveh Mollazade, Norhashila B. Hashim

Pineapple, because of its unique aroma and sweet taste, is one of the most demanded fresh-cut tropical fruit. The quality of fruit flesh is highly depend ing on its maturity level and postharvest storage conditions. Furthermore, fruit quality shows spatial differences within the fruit according to the subsequent ripening of individual fruit from bottom to top according to the appearance of the former inflorescence. Therefore, for uniform quality of fresh-cut produce , the postharvest industry requires non-contact imaging units to analyse the quality of fresh-cut pineapple. In this study, chemical properties of fresh-cut pineapples (n = 60) were evaluated by push-broom hyperspectral imaging setup in the range 380-1680 nm. A slice was cut from the middle of each fruit along the stem axis . Immediately, spectral hypercube of the slice was recorded. Then, using a cork borer, 10 cylindrical-shape samples were extracted from different locations on each slice. From the cylinders (n = 600) the chemical properties ( soluble solids content, acidity, moisture content, and caroten oids) were analysed . I mage registration was carried out to locate the samples (ROIs) in the hyperspectral images. The average spectra of the points in each ROI was considered as the spectral signature of each sample. Principal component analysis was used to compress the preprocessed spectra and consequently to reduce the size of input vector for modeling. Supervised multivariate regression analysis was applied to develop quantitative prediction models for the chemical properties of pineapples. Results showed significant differences of chemical properties between and within the fruit slices . Furthermore, hype r spectral imaging coupled with multivariate methods demonstrated huge potential for non-contact evaluation of fresh-cut pineapple (R > 0.90) produce .