Color's measurement and discrimination, of a cosmetic product by an artificial vision system

A system of artificial vision (SVA) is presented, which measures the color of a cosmetic product in the RGB coordinates. The results obtained by the SVA, were compared against the sensory evaluation of color discrimination, by an untrained panel, using statistical analysis. It was found that the SVA is able to discriminate the color of the product as equivalent to the average customer. Then, the RGB data obtained were processed using an artificial neural network (ANN) to transform them into CIE XYZ coordinates, which approximate the results of a spectrophotometer with illuminant D65. Color measurements were compared with the indicator color difference AE∗. The color measurements made by the SVA together with RNA are approximated to the spectrophotometer, however, the degree of error present, is detectable to the human visual system (HVS). The basis for the development of a research is presented, aimed at decreasing the measurement error of the system presented compared against a spectrophotometer.