Reflection modeling of rough metal surfaces using statistical theory

Bidirectional reflection distribution function (BRDF) models can accurately capture the reflective properties of target surfaces, thus playing a crucial role in achieving realistic image rendering in the field of computer graphics. The reflection model based on microfacet theory, which relies on a single roughness parameter, inadequately simulates the reflection properties of rough metal surfaces, thereby limiting rendering realism. To address this issue, we propose an improved method that introduces a glossiness parameter into the microfacet theory, thereby constructing the roughness-glossiness-Cook-Torrance (RGCT) reflection model. The glossiness parameter accurately captures the ratio and distribution of specular and diffuse reflections, which is directly related to the glossy properties of metal surfaces, enhancing the accuracy of the model in simulating the interaction between light and rough metal surfaces. We conducted comparative experiments on rough metal surfaces in the near-infrared wavelength bands with incidence angles from 0 to 90 deg. Experimental results demonstrate that compared to the traditional Cook-Torrance (CT) model based on microfacet theory, the RGCT model reduces the root mean square error (RMSE) by 0.0041 and 0.0048 on average, improves the coefficient of determination R2 by 0.0649 and 0.1082 on average, and reduces the relative error by 8.23% and 15.84% in the 1064 and 1550 nm wavelength bands, respectively. This improvement enhances the accuracy of the model in simulating the interaction between light and rough metal surfaces.