Approach for non-destructive pigment analysis in model liquids and carrots by means of time-of-flight and multi-wavelength remittance readings Original Research Article

Non-destructive spectroscopy in the visible and near infrared wavelength range has been introduced for analyzing absorbing compounds in fruit and vegetables. A drawback of the method appears due to the measuring principle, where photons detected in the diffusive tissue are influenced by the sample absorption but also scattering properties leading to variation in the photon pathlength. In the present work, distribution of time-of-flight reading was used to calculate the effective pathlength between source and detector. Using this information in addition to the spectral intensities obtained with common continuous wave spectroscopy, Lambert–Beer law was applied for analyzing absolute pigment contents. The method was tested for liquid phantoms mimicking the optical properties of fresh fruit and vegetables. Lambert–Beer law using a constant pathlength as well as combined application of the intensity at a specific wavelength and the effective pathlength resulted in low calibration errors with r2 > 0.98. Applying the two calibrations on phantoms mimicking changes in the scattering properties resulted in validation results of r2 = 0.47 and 0.64, respectively. Improved results by using the effective pathlength were confirmed on real-world samples. The carrot carotenoids analysis resulted in validation results of r2 = 0.66 and 0.74, respectively, while the measuring uncertainty was reduced from 18.10 to 9.62%. Multivariate calibrations using the entire carrot spectra and data pre-processing aiming the reduction of scattering effects resulted in slightly lower measuring uncertainty by comparison. In the sensor fusion approach proposed, however, no expensive spectrophotometer is required and the phenomenon of varying optical properties of the sample is characterized.