AUTOMATED EVALUATION OF PAINTED SURFACE QUALITY

Human aesthetic evaluation of a finished painted surface, such as an automotive body, is automatically simulated by analyzing a distorted reflected test pattern from the finished surface. Fourier power spectra of the distorted test pattern are generated to form a parameterization of the reflected distorted test pattern. A correlation is statistically made to reference Fourier power spectra, each of which has been associated with a specific human aesthetic evaluation, such as wetness, dullness, softness, brightness, and the like. The Fourier power spectra of the distorted reflected image are automatically generated through a digitized two dimensional pixel map, which are mapped into functions of the edge direction and edge intensity of the pixels. The functions are then Fourier transformed into Fourier power spectra and correlated with reference reflected patterns. Alternatively, the Fourier power spectrum of the distorted reflected pattern is optically generated by use of collimated coherent light in a Fourier transform lens. After that, the necessary parameterization is extracted from the optical Fourier transform plane and correlated to previously recorded aesthetic human evaluation. A segmentation parameter is also used for correlation with a reference segmentation indicator.