Highlights Detection for a Single Gray-Scale Image Based on Surface Shape

The existence of specular highlights is a great obstacle of shape-from-shading (SFS). For a single gray-scale image with only intensity information, the existing highlights detection methods based on chroma or polarization analysis can not directly be applied to it. So, a new method using surface shape is provided. Firstly, it makes full use of the imaging process to estimate the surface normal. Secondly, based on the physical illumination model, diffuse and specular reflection components are calculated by minimizing the error function of brightness through the simulated annealing, then locate the highlights areas by setting a threshold-value. Finally, an illumination-constrained inpainting method based on the assumption of curvature continuity is provided. Surface shape can be originated by using different methods, but the iterations will be affected by it. The physical illumination model is complex than the geometrical model, so computing will cost more time for a more accurate estimation. The constrained inpainting method will be started along the boundary of specular reflection area. It will be seen that the inpainting direction can be changed from horizontal to vertical. The experimental results show that the proposed algorithm has good stability in synthetic and real-world images, improves the accuracy of surface recovery for image combined specular highlights.