3D Scanning Of Foods For Density Calculation

Willem DE KEYZER, University College Ghent, Belgium
VAN DER SMISSEN B. 2 , DE MAEYER M. 3 , BELLEMANS M. 3 , DE HENAUW S. 3,1 , HUYBRECHTS I. 4

1 Bio- and food sciences, University College Ghent, Ghent, Belgium

Purpose
Foods, diets and nutritional status are important determinants of non-communicable diseases like cancer. Therefore, nutritional surveillance is important to monitor population intakes. During conversion of reported foods to consumed foods, coefficients for conversion are essential for proper intake estimation. When converting food volumes to weights, densities of foods are required. Density determination of foods typically associated with bulk density are challenging because the way they appear as consumed is difficult to maintain during volume measurement (for instance, salads and crisps or fries). Therefore, a method is developed to measure their volume using 3D scanning.

Methods
Foods are placed in or on a reference plate with known dimensions and volume. A 3D handheld  scanner is moved around the food while real-time surface alignment gives a good understanding of what has been scanned so far and what still needs scanning. As many scans as needed to capture the whole object can be made. During image processing scans are first aligned to get the full model. Next, all scans are fused together to get a single triangulated mesh of one 3D model. The model can then be imported into imaging software to determine the volume of the food. Together with the weight, this allows for calculation of density.
 
Results
The method is still under development and first results for salads will be presented during the conference. Also, results on accuracy and precision of the 3D scans will be presented. Previous evidence already showed the potential of 3D scanning in many different fields. 
 
Conclusions
3D scanning of foods is a potential solution for density measurements of foods with bulk density which are difficult to measure correctly using traditional underwater weighing methods.
 
Funding
Fund for Applied Research of University College Ghent