Relationship between optical properties and internal quality of melon tissues during storage and simulation-based optimization of spectral detection in diffuse reflectance mode

The optical measurement of fruit quality and the insight of light propagation in intact fruit are important to understand the optical detection mechanism and optimize the signals obtained by spectroscopy methods such as diffuse reflectance. To this end, this study measured the optical properties (absorption coefficient (mu a) and reduced scattering coefficient (mu s')) of melon tissues (exocarp, green mesocarp, and pulp) and the internal qualities (soluble solids content (SSC) and moisture content) for two cultivars ('Golden' and 'Huanghemi') during storage. The mu a spectra showed five distinct absorption peaks near 481, 675, 984, 1195, and 1423 nm for exocarp, green mesocarp, and pulp, while increasing storage time led to decreasing mu a values of melon tissues. Among the observed tissues, the exocarp had the highest scattering level in the observed wavelength ranges. The mu s ' values of the pulp decreased with the increasing storage time, while these of the exocarp and green mesocarp first decreased and then increased. The mu a and mu s ' spectra of melon pulp were both positively correlated with moisture content and negatively correlated with SSC. Next, the light propagation in intact melon fruit and optimization of spectral detection in diffuse reflectance mode were investigated by voxel-based Monte Carlo simulations based on the estimated optical properties (OPs) of melon tissues. The contribution of pulp to the detected signals increased gradually with the increase of source-detector distance, while the detected optical diffuse reflectance decreased exponentially. At the source-detector distance of 2-20 mm, 95% of the information about the pulp carried by the detected signals came from the narrow range of pulp with a depth of 7.5-12 mm and a radial width of 10.5-23.5 mm. The measured diffuse reflectance profiles of the detection configuration selected based on the simulation results followed the same trend as the simulated diffuse reflectance profiles. The optimized diffuse reflectance detection obtained considerable results for predicting SSC (Rp = 0.86, RMSEP = 0.79%, RPD = 1.93) and moisture content (Rp = 0.87, RMSEP = 0.76%, RPD = 2.06) by established partial least squares regression (PLSR) models. Accordingly, this study provides a reference for understanding the relationship between optical properties and internal quality of melons, supports a better insight into the interaction of light with the melon tissues, and helps the development of more effective non-destructive testing methods for melon fruit.