Passive reflectance sensing and digital image analysis for assessing quality parameters of mango fruits

- Actual methods for assessing mango fruit quality require numerous observations.
- Proximal remote sensing to assess quality parameters was tested.
- Models of PLSR analysis improved prediction of quality parameters.

Actual methods for assessing mango fruit quality are generally based on biochemical analysis, which leads to the destruction of fruits and is time consuming. Similarly, for valuating large quantities of mango fruits for export, numerous observations are required to characterize them; such methods cannot easily account for rapid changes in these parameters. The aims of this study to test the performance of hyperspectral passive reflectance sensing and digital image analysis was tested at various ripening degrees of mango fruits to assess their relationship to biochemical parameters (chlorophyll meter readings, chlorophyll a, chlorophyll b, total chlorophyll t, carotenoids, soluble solids content and titratable acidity) via simple linear regression and partial least square regression (PLSR) analysis. Models of PLSR included (i) spectral reflectance information from 500 to 900 nm, (ii) selected spectral indices, (iii) selected RGB indices from digital image analysis, and (iv) the combination of spectral reflectance indices and RGB indices information. The results showed that the newly developed index (NDVI-VARI)/(NDVI-VARI) showed close and highly significant associations with chlorophyll meter readings, chlorophyll a and chlorophyll t, with R2 = 0.78, 0.71, and 0.71, respectively, while the normalized difference vegetation index (Red − Blue)/(Red + Blue) was highly significantly related to chlorophyll b, carotenoids, soluble solids content and titratable acidity, with R2 values of 0.57, 0.53, 0.57, and 0.59, respectively. Calibration and validation models of the PLSR analysis based on the combination of data from six spectral reflectance indices and six RGB indices from digital image analysis further improved the relationships to chlorophyll meter readings (R2 = 0.91 and 0.88), chlorophyll a (R2 = 0.80 and 0.75), chlorophyll b (R2 = 0.66 and 0.57) and chlorophyll t (R2 = 0.81 and 0.80), while calibration and validation models of PLSR based on the data from the spectral reflectance range from 500 to 900 nm were most closely related to soluble solids content (R2 = 0.72 and 0.48) and titratable acidity (R2 = 0.64 and 0.49). In conclusion, the assessment of biochemical parameters in mango fruits was improved and more robust when using the multivariate analysis of PLSR models than with previously assayed normalized difference spectral indices and RGB indices from digital image analysis.