Back to Programme
A11-P: OXIDATIVE PROCESS IN BASIL-BASED "PESTO" SPREADS – COLOUR EVALUATION
L. Pezo, M. Pavlović, S. Ostojić, M. Kićanović, S. Zlatanović, O. Kovačević,
B. Simonović
Institute of General and Physical Chemistry, Beograd, Studentski trg 12/V, Serbia
Colour evaluation, by means of multivariate image analysis was applied on colour lightness changes, corresponding to enzymatic and non-enzymatic browning of mild-heat processed basil-based food emulsions. The colourgram of digitalized images, taken during oxidation of samples, was evaluated. Basic colour information was derived from frequency colour distribution for RGB (red, green and blue) and HSV (hue, saturation and value) colour system, while the Principal Component Analysis (PCA) was carried out by evaluating covariance matrix, for raw, mean centered and autoscaled matrix of the basic R, G, B colour information, and the evaluation of eigenvalues and eigenvectors for all three numeric models. Influence of metal-chelating protein lactoferrin and organic acids on changes in lightness frequency distribution (L), corresponding to browning, was determined.
Colour images of basil-based pesto spreads were captured by a Sony PowerShot A550 CCD camera for multivariate colour analysis, applied here for representing the colour lightness changes due to oxidative browning of minimally-processed basil-based food emulsions. The influence of metal-chelating protein lactoferrin was also analyzed on browning of emulsions (atributed to enzymaic and non-enzymatic oxidation). The evaluation of colourgrams of digitalized images, which were taken during oxidation of the samples, was done during experiment. The colour information was recorded for each pixel and the derived frequency colour distribution for RGB colour system were examined, while the PCA analysis was carried out by evaluating covariance matrix, for raw, mean centered and autoscaled matrix of the R, G, B colour distribution. The evaluation of eigenvalues and eigenvectors for all three numeric models was also performed.
The relative colour values were also calculated, and they were also considered since a particular colour, e.g. red colour does not correspond to a high absolute value of the corresponding R variable, but to a high value of the ratio between R and L.The hue, saturation and intensity (HSV) values are also calculated, for the conversion from the RGB to the HSV colour space.
The principal component analysis (PCA) was performed using the score matrix of raw, mean centered, and autoscaled unfolded RGB distribution matrix. Mean centered matrix was evaluated by evaluating the subtracting the mean value of each color vector to each element of the vector, and the autoscaled matrix was evaluated by dividing the mean centered matrix vectors with standard deviations for each vector. The evaluation of covariance matrix for all three models matrix followed, with the evaluation of eigenvectors and eigenvalues of covariance matrix. The eigenvectors of covariance matrix for raw, mean centered and autoscaled were treated as loading vectors in PCA analysis. The eigenvectors of covariance matrix are then calculated, for all three models. Finaly, the score matrix is evaluated for all three models. After evaluation of all these data, colourgram was ploted.
References
- A. Antonelli, M. Cocchi, G. Fava, G.C. Franchini, D. Manzini, A. Ulrici, Anal. Chim. Acta, 2004, 515, 3-13.
- J. Huang, H. Wium, K.B. Qvist, K.H. Esbensen, Chemometrics Intell. Lab. Syst., 2003, 66, 141 – 158.
