Metal Surface Texture Reconstruction Based on Near-Field Photometric Stereo

The surface quality of precision metal parts will affect the performance and appearance of the product. For this reason, it is necessary to measure and evaluate the surface texture of the parts. Commonly used contact measurement methods have low measurement efficiency, and optical measurement methods are susceptible to the effects of surface highlight reflection. For this reason, a method based on near-field non-Lambertian photometric stereo vision for highlight metal surface texture reconstruction is proposed. In order to effectively describe the non-Lambertian reflection, this method uses an inverse reflectance model based on an inverse reflectance light source to decouple the surface normal vector and the nonlinear reflection model. This method uses neighborhood information to improve the robustness of the inverse model, and uses the maximum fusion strategy to overcome the influence of shadows and rendering generates targeted simulation datasets, thereby improving the adaptability to metal surface reflection. The results show that the proposed method can reconstruct the high-brightness metal surface texture with high precision, and the relative measurement error is less than 15%.