Green Banana Maturity Classification and Quality Evaluation Using Hyperspectral Imaging

Physiological maturity of bananas is vital importance in determination their quality and marketability. This study assessed, with the use a Vis/NIR hyperspectral imaging (400–1000 nm), feasibility differentiating six levels (maturity level 2, 4, 6 to 9) green dwarf banana characterizing changes during maturation. Spectra were extracted from three zones (pedicel, middle apex zone) each finger, respectively. Based on spectra zone, identification models high accuracy (all over 91.53% validation set) established by partial least squares discrimination analysis (PLSDA) method raw spectra. A further generic PLSDA model an 94.35% for was created zones’ pooled omit effect acquisition position. Additionally, spectral interval selected simplify model, built 85.31% set. For some main parameters (soluble solid content, SSC; total acid TA; chlorophyll content chromatism, ΔE*) banana, full-spectra (PLS) PLS were, respectively, developed correlate those data. In models, coefficients (R2) 0.74 SSC, 0.68 TA, fair 0.42 as well 0.44 ΔE*. The performance slightly inferior that models. Results suggested SSC TA had acceptable predictive ability (R2 = 0.64 0.59); ΔE* 0.34 0.30) could just be used sample screening. Visualization maps also applying pixel image, distribution which basically consistent actual measurement. proved useful tool assess bananas.