CV equipment Database

Color-difference inspection of household appliance surfaces over large fields of view and across different materials faces pronounced accuracy bottlenecks. Variations in refractive index, surface roughness, and gloss markedly change the reflection and scattering of incident light, so single-reference color matching cannot maintain consistency across materials and yields systematic bias for \u201cidentical color, different materials\u201d. This paper proposes a high-precision appliance surface color-difference method based on cross-material optical modeling. Starting from the imaging mechanism, flat-field and illumination compensation are performed to correct the spatial dependence of luminance and chromaticity, followed by linear color reconstruction and sRGB mapping to recover an equivalent standard illuminant. A lightweight bidirectional reflectance distribution function prior is then introduced to compensate perceptual inconsistency caused by directional effects of roughness and gloss. On this basis, a material-consistency correction factor and polynomial chromaticity mapping are constructed to form a unified cross-material color space, together with an angle-weighted improved color-difference metric. Experiments on a 2 m \u00d7 2 m field of view with 350 samples (50 colors, 7 material types) show that the proposed method reduces the mean color difference from 4.49 for raw RGB readings and 3.57 for simple color conversion to 1.25, an overall reduction of about 72%. The board-level is stabilized around 1, and the consistency index increases from 0.75 to 0.94, close to the 0.98 of a professional colorimeter, while effectively suppressing pseudo color differences due to \u201csame color, different materials or gloss\u201d.