Quantification of perceptual coarseness of metallic coatings containing aluminum flakes using texture analysis and visual assessment methods

Abstract

The industrial need for controlling the quality of metallic coatings for producing efficient and high-quality products makes it necessary to conduct an appearance quantification process. In this study, it had been attempted to present an appropriate algorithm to analyze the coarseness property of the metallic coatings using the image processing technique and texture analysis methods to achieve acceptable correlation with visual perception. For this purpose, we examined metallic silver and metallic blue coatings containing eight grades of aluminum flakes with different size distributions from fine to coarse cases. After capturing the images of samples using a scanner, they were studied using different texture analysis methods including autocorrelation function, distance-dependent edge frequency, gray level co-occurrence matrix (GLCM), Fourier transform and wavelet transform. Visual assessments were performed using a comparative judgment method. The precisions of the applied analysis methods were evaluated based on the correlation between their results and the visual assessment data. According to the obtained data, the autocorrelation function with R-squares of 0.88 and 0.9 for the metallic silver and metallic blue coatings, and also wavelet transform in the vertical channel with R-squares of 0.86 and 0.89 for the same mentioned coatings, respectively, had the best performance in comparison to visual perception. Consequently, it was possible to precisely quantify the coarseness perception of metallic coating using autocorrelation function and the wavelet transform as two feasible texture analysis methods. Moreover, the presence of the blue absorption pigment in the formulation had no significant effect on the performance of these two methods.