基于控制点约束影像的密集匹配及其在考古发掘中的应用

Based GCP-SGM Algorithm and Its Application in Scene 3D Reconstruction of Archaeological Excavation Site

针对大场景考古挖掘现场的三维重建情况,选取半全局匹配策略作为研究基础,设计了一种基于附加控制点约束的半全局匹配算法进行序列影像密集匹配。将稀疏匹配中的特征提取和匹配技术用于提取初始特征,并由这些初始特征转化的同名特征点,生成视差空间影像中的视差控制点,以此作为一种可靠的约束,提高其密集匹配的精度。同时,采用影像分块的策略,将原始大核线影像分成若干对小核线影像,进行密集匹配,以达到提高计算效率,改善计算结果的目的。实验证明,该方法可实现大面积考古挖掘现场的快速三维重建,并且能够在进行文物形态三维重建的同时获取挖掘现场文物分布的正射影像平面图,为准确记录挖掘现场文物分布位置及考古发掘、调查及遗址保护规划编制等提供科学依据。

Aiming at the big scene 3D reconstruction of archaeological excavation site,in the paper,a method which based on photogrammetry and computer vision is put forward with Qinshihuang＇s acrobatics figurines pits archaeological excavate as example.Getting multi-strip sequence images usednonmetric digital camera,based on the control points laying the bottom of the excavation site,the exterior orientation elements of the image is measured accurately and the feature matching tie points of the images and epipolar image can be acquired by aerial triangulation.On the basis,the SGM（semi-global matching）algorithm with the additional control point constraint is used for stereo dense image matching.The per-pixel dense matching results are calculated by space intersection.at the same time,the dense color three-dimensional point cloud which reflects the cultural relics archaeological excavation site location information is obtained,and to achieve the 3Dreconstruction of cultural form in archaeological excavation process.Experimental results show：the method given by this paper can achieve rapid measurement of 3Dreconstruction for the big scene of archaeological excavation site,and the orthophoto map which reflects the distribution of cultural relic site is obtained at the same time,The proposed method can provide the accurate distribution of dig site cultural relic and scientific basis in archaeological excavation,survey,site protection planning and so on for related fields studying.