This paper reports a preliminary work on a feasibility study of applying terahertz (THz) technology for pecan quality
evaluation. A set of native pecan nuts collected in 2009 were used during the experiment. Each pecan nutmeat was
manually sliced at a thickness of about 1mm, 2mm, and 3mm and a size of about 2cm (length) ×1cm (width). Pecan
shell and inner separator were also cut into the same size. The absorption spectra for the nutmeat slices, shell, and inner
separator were collected using a THz time-domain spectroscopy (THz-TDS) developed by a group of researchers at
Oklahoma State University. The test results show that nutmeat, shell, and inner separator had different absorption
characteristics within the bandwidth of 0.2-2.0 THz. To study the capability of insect damage detection of the THz
spectroscopy, the absorption spectra of insects (living manduca sexta and dry pecan weevil) were also collected. Due to
high water contents in the insects, very obvious spectral characteristics were found. The results from the preliminary
study show a potential of THz technology applied for quality detection of bio-products. However, since bio-products
mostly have high water content and are handled under an environment with certain levels of water content, practical
issues needs to be further investigated to make the THz technology a feasible tool for quality evaluation.
Meat grading has always been a research topic because of large variations among meat products. Many subjective assessment methods with poor repeatability and tedious procedures are still widely used in meat industry. In this study, a hyperspectral-imaging-based technique was developed to achieve fast, accurate, and objective determination of pork quality attributes. The system was able to extract the spectral and spatial characteristics for simultaneous determination of drip loss and pH in pork meat. Two sets of six significant feature wavelengths were selected for predicting the drip loss (590, 645, 721, 752, 803 and 850 nm) and pH (430, 448, 470, 890, 980 and 999 nm). Two feed-forward neural network models were developed. The results showed that the correlation coefficient (r) between the predicted and actual drip loss and pH were 0.71, and 0.58, respectively, by Model 1 and 0.80 for drip loss and 0.67 for pH by Model 2. The color levels of meat samples were also mapped successfully based on a digitalized Meat Color Standard.