Scanner-based color measurement in L*a*b* format with artificial neural networks (ANN)

Kihc K., Onal-Ulusoy B., Yildirim M., Boyaci I. H.

EUROPEAN FOOD RESEARCH AND TECHNOLOGY, vol.226, pp.121-126, 2007 (SCI-Expanded) identifier identifier


A computerized inspection system (CIS) that uses a flat-bed scanner, a computer, and an algorithm and graphical user interface coded and designed in Matlab(R) 7.0 was developed to determine food color based on CIE L*a*b*, a color format. The USA Federal Color Standard printouts (SP) comprised of 456 different colors were used to train and test the artificial neural network (ANN) integrated CIS. Strong orrelations were found between the results estimated from ANN-integrated CIS and those obtained from spectrophotometer (R (2), 0.991, 0.989, and 0.995 for L*, a*, and b*, respectively) for test images data set. Various food samples were also evaluated to test the performance of the CIS. A good agreement, R-2, 0.958, 0.938, and 0.962 for L*, a*, and b*, respectively, was found between color measurement with CIS and a spectrophotometer. CIS with a mean error of 0.60% and 2.34% for test and various food samples, respectively, has an ability to imitate the results obtained from a spectrophotometer. CIS allows users to store the captured picture for further use and estimate the overall color or the color of selected region of the samples either heterogeneous in color or amorphous in shape.