Multispectral Imaging of Meat Quality - Color and Texture
In DTU Compute PHD-2014, 2015
Abstract
The use of computer vision systems in food production and development is increasing. Computer vision systems offer fast, reliable, objective and noninvasive methods for assessment of wanted quality traits. This thesis investigates the applicability of computer vision systems in the assessment of meat quality parameters, especially with regards to meat color and texture. Several image modalities have been applied, all considering multi- or hyper spectral imaging. The work demonstrates the use of computer vision systems for meat color measurements. The color is assessed by suitable transformations to the CIELAB color space, the common color space within food science. The results show that meat color assessment with a multispectral imaging is a great alternative to the traditional colorimeter, i.e. the vision system meets some of the limitations that the colorimeter possesses. To mention one, it is possible to assess color of very complicated structures, such as salamis, with a vision system. More importantly though, the vision system embraces the complicated scattering properties of meat. The images can also lead to other analyses, e.g. image texture analysis relating to the structure of the meat. In the thesis it is presented how simple texture measures can be used for characterizing the texture changes in fermented salamis. Moreover, it was investigated if it was possible to relate structure in images to chemical compounds in lard from boars.
Info
Thesis PhD, 2015
In DTU Compute PHD-2014, 2015
UN SDG Classification
DK Main Research Area
Science/Technology
