Single-view 3D registration aims to estimate the position and orientation of the object in 3D space. Real-time and accurate 3D registration is one of the key technologies in augmented reality, which is a means to guarantee the geometric consistency of augmented reality system. For the lack of features of the textureless objects, single-view 3D registration method based on contour matching is a general method for 3D registration with textureless model, but it is limited to the derivative of the solution method. In this paper, we propose a new simulate calculation method for single-view 3D registration. In the case of the non-derivative function, the steady gradient is obtained by the graphical simulation method, and the stable and robust pose parameter is solved. Based on the analysis of the projection accuracy of the 3D model, the simulation results of the 3D model are analyzed. The smooth expression of the objective function is obtained according to multiple sampling of the model and depth analysis, thus the stable 3D pose result is obtained. The quantitative and qualitative experimental results show that this method can converge quickly, and can achieve very high accuracy and robustness. At the same time this method effectively solves the problem of local extremum.
Li, C., Ma, H., & Gao, X. (2018). Model-based 3D registration optimization based on graphics simulation method. In Procedia Computer Science (Vol. 131, pp. 344–353). Elsevier B.V. https://doi.org/10.1016/j.procs.2018.04.173