Evaluation of scanner capability for measuring the color of fabrics with different textures in different setups

The influence of surface texture on the perceived color by a flatbed scanner is investigated. Knitted fabrics with 8 different textures in variety of colors are prepared and used in scanning trial along with standard chart, i.e. IT8.7/2 from Kodak. According to the results, the scanner detects different RGB values for different textures which are weaved from same colored yarns. The means of R, G and B values are considered as a feature vector which shows the dominant color of each sample. Samples are scanned with different resolutions and it is found that scanning resolution does not change the extracted color feature vector. The IT8.7/2 standard target which benefits from non textural solid surface and fabrics with specific surface texture are used for colorimetric characterization of scanner and the capability of scanner for estimation of color coordinates of samples with different textures is examined. It is found that the characterization of scanner with textured target improves the scanner accuracy for the color of textured materials. It is shown that the texture of media has a great effect on the characterization results and there is relatively good correlation between the structural differences of textures of fabrics used in training and testing steps with the mean of testing color difference values. To achieve a better color reproduction results for fabrics, scanner characterization should be performed for each set of fabrics with specific texture. In this case, increasing the bit depth of captured image in the scanning process leads to decrease of the mean of color difference value for training and testing packages.