Calibrated color measurements of agricultural foods using image analysis
Introduction
Color is considered a fundamental physical property of agriculture products and foods, since it has been widely demonstrated that it correlates well with other physical, chemical and sensorial indicators of product quality. In fact, color plays a major role in the assessment of external quality in food industries and food engineering research (Segnini et al., 1999, Abdullah et al., 2001).
Color can be rapidly analyzed by computerized image analysis techniques, also known as computer vision systems (CVS). These systems not only offer a methodology for measurement of uneven coloration but it can also be applied to the measurement of other attributes of total appearance (Hutchings, 1999). So far it has not been attempted in the published literature to detect all the optical properties of an object from object images, and in particular, for those with irregular surfaces. However, calibrated average color measurements and other appearance features using computer vision techniques have been shown to closely correlate with those from the visual assessment (Mendoza and Aguilera, 2004).
Dedicated commercial vision systems are currently available for a variety of industrial applications, and they are especially recommended for color assessments in samples with curved and irregular shapes of the surface; however, the effects of these physical properties and how they are handled to give representative measurements is frequently not known. The knowledge of these effects, such as the variations of L*, a*, and b* for a particular shape of the sample, could be useful for developing image processing correction algorithms which can permit a better correlation among product quality by CVS and human vision evaluations.
The objectives of this study were: (i) to implement a computerized image analysis technique to quantify standard color according to the CIE system based on the sRGB standard; (ii) to assess the effect of changes in the commonly used capture conditions of the CVS with regard to the initial calibration conditions using L*a*b* color space; and (iii) to critically assess the sensitivity of the RGB, L*a*b* and HSV color spaces to represent the color in curved surfaces such as bananas and red pepper.
Section snippets
Color concepts
Color is basically specified by the geometry and spectral distributions of three elements: the light source, the reflectivity of the sample and the visual sensitivity of observer. Each of these was defined by the Commission Internationale de I’Eclairage (CIE) in 1931. The definition was aimed at simulating the human color perception based on a 2° field of view, a set of primaries (red, green, blue), and color-matching functions (CIE, 1986). The matching functions for the standard observer (
Calibration of the CVS: transformation of computer R′G′B′ into CIE XYZ space
Fig. 3a shows the plot of the R′G′B′ values acquired by CVS against their corresponding RGB values measured by colorimeter on the 125 Pantone® color sheets. The results clearly showed the nonlinear relationship among RGB values that is typical of many devices, meaning that the expected gamma correction needs to be applied to each of the tristimulus values of the CVS.
Fig. 3b shows that using the recommended IEC transfer functions (Eqs. (1)–(3)), which considered a gamma value equal to 2.4, a
Conclusions
A de facto standard (sRGB) for the spectral sensitivities of practical recording devices adopted by the IEC 61966-2-1 (1999) was shown to be straightforward, efficient and simple to implement in Matlab. The correlation coefficients for XYZ values obtained from calibrated CVS and colorimeter for 125 color sheets showed good color matching with an R2 > 0.97 in all regressions.
A sensitivity analysis of the implemented CVS using bananas demonstrated that the color measurements of lightness (L*
Acknowledgments
Thanks to MECESUP/PUC 9903 Project (Chile) for granting the first author a doctoral scholarship at the School of Engineering, Pontificia Universidad Católica de Chile, and ALFA Programme (EC) – EU Alfa Project II-0121-FA for financial assistance to develop part of this investigation at Lund University.
References (27)
- et al.
Inspection and grading of agricultural and food products by computer vision systems—a review
Comput. Electron. Agric.
(2002) - et al.
Improving quality inspection of food products by computer vision—a review
J. Food Eng.
(2004) - et al.
Machine vision technology for agriculture applications
Comput. Electron. Agric.
(2002) - et al.
Recent developments in the applications of image processing techniques for food quality evaluation
Trends Food Sci. Technol.
(2004) - et al.
Comparison of three methods for classification of pizza topping using different colour space transformations
J. Food Eng.
(2005) - et al.
Calibrated colour imaging analysis of food
- et al.
Evaluation of pork color: prediction of visual sensory quality of meat from instrumental and computer vision methods of color analysis
Meat Sci.
(2003) - et al.
A low cost video technique for color measurement of potato chips
Lebensm.-Wiss. U.-Technol.
(1999) - et al.
A simple digital imaging method for measuring and analyzing color of food surfaces
J. Food Eng.
(2004) - et al.
The applications of computer vision system and tomographic radar imaging for assessing physical properties of food
J. Food Eng.
(2001)