The development of food factory in the future essentially consists of three major aspects: control of contaminations, reduction of environmental impact, and optimization of production processes. In this regard, conventional vis/NIR spectroscopy has been adopted by the industry as a means for on-line/off-line non-destructive sensing of food quality or process control and proven to be a valuable approach for these aspects. This technique tackles these issues by allowing non-destructive quality assessment of individual products in batches or on production lines; which, to some extent, helps to reduce wastes or contaminants by destructive quality tests. It also provides a means for optimizing production processes by real time monitoring of production lines. In research point of view, the conventional vis/NIR spectroscopy which bases on quantifying the absorption of targeted components in the post-interacting lights at a single point to predict their concentrations in the analyzed samples reveals some downsides. Typically, the acquired post-interacting lights are the results of both intertwined phenomena: scattering, related to refractive index differences causing changes in direction of the propagating lights and thus microstructure, and absorption, related to chemical composition. Due to scattering effects, light propagation path lengths change; which consequently induces over-estimation of the absorption levels in samples having the same composition but different microstructures. Scattering has been, therefore, considered as ?noise? in vis/NIR spectroscopic data analysis and needs to be removed or reduced for improved prediction performance. However, in food quality point of view, besides chemical composition, scattering also contains information about food microstructure which is also related to many important quality attributes such as sponginess of bread, crispiness or crunchiness of crackers, firmness of fruits, etc.

This research, therefore, aims at investigating the potential of spatially resolved spectroscopy (SRS) for acquiring both chemical composition and microstructure of food. By combining multiple conventional vis/NIR spectroscopic measurements at different source-detector distances (spatially resolved), scattering and absorption properties of the samples can be separated with light propagation models. A setup for SRS was elaborated by combining a fiber-optics probe with a spectrograph-camera combination for measurement in the 400-1000 nm range.  The probe consists of seven optical fibers: one serves as illumination fiber and the others serve as detection fibers to collect diffuse reflected light at different source-detector distances ranging from 0.3 to 1.2 mm. This probe has been used to characterize sugar foams, having the same chemical composition, prepared under different processing conditions (foaming times) to create different air bubble size distributions in them. These different microstructures were expected to produce different scattering properties of light thanks to the mismatch in refractive indices at the bubble interfaces. Preliminary results have shown significantly different reduced scattering coefficients in the range 400-1000 nm of 3 sugar foams prepared with foaming times of 1, 5, and 10 minutes; which indicated that different bubble size distributions in these foams, or different microstructures, have been successfully differentiated non-destructively by SRS technique.