|Qin, Jianwei - MICHIGAN ST UNIVERSITY|
Submitted to: Proceedings of SPIE
Publication Type: Proceedings
Publication Acceptance Date: November 15, 2005
Publication Date: December 15, 2005
Citation: Qin, J., Lu, R. 2005. Hyperspectral diffuse reflectance for determination of the optical properties of milk and fruit and vegetable juice. Proceedings of SPIE. 5996:59960Q. Interpretive Summary: Machine vision and spectroscopy are being widely used for quality evaluation of horticultural and food products. These optical techniques generally rely on measuring and analyzing light reflected or transmitted from product items for assessing their quality attributes, either external or internal. Most food products are turbid and their optical properties are characterized by the spectral absorption and reduced scattering coefficients. Measurement of these optical properties is important for understanding light interaction with food materials and designing effective optical inspection systems. However, little is available in literature on these optical properties for horticultural and food products. Current measurement techniques are generally expensive, time consuming and restrictive in sample preparation, and hence they are not suitable for horticultural and food products. This paper reports on the development of a new method and technique for fast determination of the optical properties of food products. A hyperspectral imaging system, which acquires both spectral and spatial information from an object in the visible and near-infrared (longer than the visible) spectral region, was assembled for measuring the optical properties of food materials. A calibration procedure coupled with a theoretical model was developed for extracting optical properties from the hyperspectral reflectance images of turbid foods. The optical properties of a variety of fruit and vegetable juices and milk were determined over the spectral range between 530 nm and 900 nm. It was found that absorption and reduced scattering coefficients had a high correlation with the fat content of milk. The technique is fast, nondestructive or noninvasive and relatively easy to implement, and it can be used for determining the optical properties of a large variety of horticultural and food products. Establishment of such an optical properties database will help researchers and instrumentation engineers in designing better optical systems for quality evaluation of horticultural and food products.
Technical Abstract: Absorption and reduced scattering coefficients are two fundamental optical properties for turbid biological materials. This paper presents the technique and method of using hyperspectral diffuse reflectance for fast determination of the optical properties of fruit and vegetable juices and milks. A hyperspectral imaging system was used to acquire spatially resolved steady-state diffuse reflectance over the spectral region between 530 and 900 nm from a variety of fruit and vegetable juices (citrus, grapefruit, orange, and vegetable) and milks with different fat levels (full, skim and mixed). The system collected diffuse reflectance in the source-detector separation range from 1.1 to 10.0 mm. The hyperspectral reflectance data were analyzed by using a diffusion theory model for semi-infinite homogeneous media. Absorption and reduced scattering coefficients of the fruit and vegetable juices and milks were extracted by inverse algorithms from the scattering profiles for wavelengths of 530-900 nm. Values of the absorption and reduced scattering coefficient at 650 nm were highly correlated to the fat content of the milk samples with the correlation coefficient of 0.990 and 0.989, respectively. The hyperspectral imaging technique can be extended to the measurement of other liquid and solid foods in which light scattering is dominant.