Non-destructive measurement of mandarin orange quality
VI.P8 Non-destructive measurement of mandarin orange quality parameters with visible near infrared spectroscopy L. Paulo, M. Resende, A. Nunes , C. Miguel Pintado C, P. Antunes CATAA- Associação Centro de Apoio Tecnológico Agro-Alimentar de Castelo Branco, Zona Industrial de Castelo Branco, Rua A, 6000-459 Castelo Branco, Portugal. Introduction The development of sensors to measure fruit internal quality variables is one of the challenges of post-harvest technology. Visible-near infrared spectroscopy (Vis/NIRS) has been a promisable technique for non-destructive fruit quality assessment. This study was focused to evaluate the use of Vis/NIRS in measuring the quality parameters of intact mandarin orange fruit (Citrus clementina). Material and Methods • Vis/NIRS - Spectra were collected with a contact probe (ASD Labspec 4.0 NIR) • Fruit colour, L*, a* and b* (Minolta CR-400) • Firmness (TA.XT Plus, Stable Microsystems) Compression until 5 % deformation • Soluble solids content, SSC (Atago PR32α) • Titratable acidity, TA (Titromatic 2S -Crison) Results Parameter Minimum Maximum Parameter L* a* 45.77 71.82 Minimum SSC (%) 8.20 Maximum 11.6 pH b* -10.24 41.99 33.62 69.74 peel thickness (mm) 1.6 5.8 3.2 TA (mEq. 100 g-1) 9.47 Firmness (N) 4.84 3.9 17.5 14.29 Parameter L* a* b* peel thickness SSC pH TA Firmness L* a* b* peel thickness SSC TA pH Firmness Calibration preprocess Min-max Min-max SNV SNV SNV Min-max 1der+SNV 2der Validation preprocess Min-max Min-max SNV SNV SNV 1der+SNV Min-max 2der Rank 3 4 4 7 7 7 r2 0.923 0.963 0.852 0.909 0.749 0.535 RMSEC 0.876 1.780 1.580 0.382 0.314 0.095 Bias -4.54e-8 8.51e-7 -4.09e-7 -1.56e-8 -2.72e-7 -7.95e-8 7 5 0.629 0.650 1.101 1.030 -4.77e-7 1.14e-7 Rank 3 4 4 7 7 7 7 5 r2 0.939 0.971 0.893 0.791 0.747 0.648 0.517 0.516 RMSEP 0.871 1.803 1.536 0.625 0.312 1.073 0.094 1.536 Bias -0.066 0.192 -0.218 0.082 -0.019 0.044 0.016 -0.136 Rank, number of factors used in prediction models; r2, coefficient of determination; RMSEC, root mean square error of calibration; RMSEP, root mean square error of prediction; Bias, estimate of the test error at that same value of the tuning parameter; 1der, first derivative; 2der, second derivative; SNV, vector normalization; Min-max, minimum-maximum normalization. Conclusions The models presented a good prediction for color parameters, peel thickness, and SSC. pH and acidity showed low variability in the samples reducing the correlation and accuracy of the model. The results indicate that Vis/NIRS technique could provide an accurate, reliable and nondestructive method for assessing the internal quality indices. Acknowlegments This research was supported by MITTIC. The MITTIC Project, Technological Modernization and Innovation based on ICT in strategic and traditional sectors, is financed jointly by the European Regional Development Fund (ERDF), through the Operational Programme of Cross-border Cooperation Spain - Portugal (POCTEP) 2007-2013.