Sequential fusion of information from two portable spectrometers for improved prediction of moisture and soluble solids content in pear fruit

Near infrared (NIR) spectroscopy allows rapid estimation of quality traits in fresh fruit. Several portable spectrometers are available the market as a low-cost solution to perform NIR spectroscopy. However, spectrometers, being lower cost than benchtop counterpart, do not cover complete near spectral range. Often sensors either use silicon-based visible and detector 400–1000 nm, or InGaAs-based short wave (SWIR) covering 900–1700 nm. these two regions carry complementary information, since nm interval captures color 3rd overtones most functional group vibrations, while 1st 2nd same transitions fall 1000–1700 To exploit such complementarity, sequential data fusion strategies were used fuse from i.e., Felix F750 (~400–1000 nm) DLP Scan Nano (~900–1700 nm). In particular, different approaches used, namely orthogonalized partial-least squares (SO-PLS) regression covariate selection (SO-CovSel). SO-PLS improved prediction moisture content (MC) soluble solids (SSC) pear fruit, leading an accuracy which was obtainable with models built on any set individually. Instead, SO-CovSel select key wavelengths both ranges mostly correlated parameters Sequential led model (higher R2 RMSEP) MC SSC fruit: compared (the worst individual block) where showed increase R2p up 56% corresponding 47% decrease RMSEP. Differences less pronounced alone, but still increased by 2.5% RMSEP reduced 6.5%. is limited it can be considered general tool for integrating information multiple sensors.