92D-4 |
Correlation of color measurement between a color machine vision system and a hand-held colorimeter |
A. Z. ODABASI and M. O. Balaban. Dept. of Food Science & Human Nutrition, Univ. of Florida, 359 FSHN Bldg., Newell Dr., PO Box 110370, Gainesville, FL 32611-0370 Color quantification using machine vision is becoming more accessible, with recent advances in digital imaging. Color machine vision (CMV) has advantages over the instruments of colorimetry and spectrophotometry: It allows a larger area to be sampled; even the entire sample surface can be analyzed. Samples with non-uniform color and/or surface that are misrepresented in conventional instruments can be evaluated objectively in this high-spatial resolution method. Since it is a non-contact, non-destructive method, temporal color changes in a given sample can also be quantified. The objective of this study was to compare and correlate the measurements of a CMV system and a colorimeter. The CMV system consisted of a light box and a CCD color camera connected to a computer with frame grabber. Using the previously developed software, images of samples were grabbed to obtain color information. As soon as the sample was removed from the light box, a Minolta CR-200b colorimeter was used to obtain the L*a*b* values. The samples were marked to minimize the probability of reading different locations during each measurement. Tiles and peppers of four different colors were used as samples. A high correlation (R-square >0.97) was obtained among the L*a*b* measurements of the two methods when standard color tiles were used. When peppers were sampled, however, the non-uniformity of the sample color not detected in the colorimeter due to its small sampling port caused discrepancies in the two sets of data. This was expected since the CMV system used a larger area on the non-uniform pepper samples that increased variability. Correlation equations and how well they perform will be presented. CMV is a promising tool in both off-line and on-line color evaluation of foods. A good understanding of how such a system’s measurement relates to that of a conventional instrument is crucial.
Session 92D, Quality Assurance: General
|