Three-dimensional Reconstruction Of Terracotta Warriors From Multi-angle Images

The magnificent Qin Shihuang Terracotta Warriors,a typical crystallization of the wisdom of the ancient Chinese people,marks the level of war culture and craftsmanship in the historical period,full of the impressive artistic charm,attracting people to deeply explore the ancient Chinese plastic arts.After decades of archeological excavation,conservation research,and with the continuous development of computer vision research,the approaches about the protection of the Terracotta Warriors tend to be digitized.This paper mainly focuses on the methods about the 3D reconstruction from the multi-angle images of the Terracotta Warriors.The main research contents are as follows:(1)According to the high mismatch rate and low efficiency in 3D reconstruction of the Terracotta Warriors,a novel feature partition matching scheme on multi-angle images is proposed.Experimental results show that about the terracotta image feature extraction and matching,the correct matching rate in the new scheme can reach 96.8%,and the new scheme also has better robustness when scale,illumination and angle are changed.Thus,the scheme can match the feature points with greater precision,greatly improving the applied value in the 3D reconstruction of the Terracotta Warriors.(2)Based on the(1)research,an improved SFM(Structure From Motion)method is proposed to achieve the 3D reconstruction of multiple Terracotta Warriors images.Compared with the original SFM method,the improved method,combining the SFM method with the feature partition matching scheme,reconstructs a more complete sparse point cloud model.(3)Considering that the sparse point cloud is too different from the actual target object,it is difficult to express the 3D information of the reconstructed object.Based on the(2)research,a PMVS(Patched-based Multi-View Stereo)method is used to get a more realistic,colored dense point cloud of the Terracotta Warriors.Compared with the dense point cloud obtained by the original SFM method,the method can show the real 3D information of the reconstructed target with greater accuracy and precision.(4)Based on the above research,this paper designs and implements a three-dimensional reconstruction system for Terracotta Warriors.The system functions include the system include image feature extraction,feature partition matching,sparse reconstruction and dense reconstruction of Terracotta Warriors.The system can effectively reconstruct the three-dimensional model of the Terracotta Warriors,and it also verifies the effectiveness of the method.