Improved algorithm for estimating the optical properties of food products using spatially-resolved diffuse reflectance

Understanding the optical properties of food products is essential to apply optical methods for quality detection. In this research, the inverse algorithm for estimating the optical properties of food products from spatially-resolved diffuse reflectance was optimized based on the reflectance profiles generated by Monte Carlo simulations. First, the start and end points, namely, the spatial region, of reflectance profiles for parameter estimation were optimized with the unit of mean free path (mfp'). Based on this optimal spatial region with recommended start and end points, an improved step-by-step method was proposed over conventional 1-step (C-1-step) method that used fixed start and end points. Furthermore, the C-1-step method was also modified by using the optimized end point of 16 mfp's and designated as M-1-step method. Results showed that the optimal start point decreased from 3 mfp's to 1.6 mfp's with the increase of μs′/μa value from 10 to 60, while the recommended end point was kept at 16 mfp's. Absolute values of relative errors for the best estimates of μa and μs′ using C-1-step and step-by-step methods were 8.7%, 5.6% and 3.5%, 2.3%, respectively, for a spatial resolution of 0.1 mm. The M-1-step method, with the spatial resolution of 0.1 mm, reduced the estimation errors to 3.8% and 3.7% for μa and μs′, representing 56% and 34% improvements over the C-1-step method. Based on the results of M-1-step method, the step-by-step method could also improve estimation accuracy. At last, the effectiveness of the proposed algorithm was validated with real mango samples. The mean absolute percentage errors for estimating μa and μs′ by a hyperspectral imaging system combined with the step-by-step method were 9.2% and 5.7%, respectively.