Point Cloud Data Processing Method For Archaeological Excavation 3D Scene Reconstruction

Digital Archaeology is an important part of digital earth,field archeology as the basic method to obtain first-hand data for Archaeological research,it has the characteristics of dynamic,hierarchical mining,stage results and other characteristics.Recording every stages of the unearthed relics and remains timely and accurately is the necessary work in archaeological recording.In the past,we only using a single stage of collecting data to 3D reconstruction,it is hard to completely show the various periods original scene of Archaeological Sites,and construct the 3D model of the site.However,collecting data continuously to excavation sites is meet the basic requirements for following the specification for Field Archaeology and completely recording all kinds of remains.Three dimensional laser scanning is a kind of spatial data acquisition method,which has been developed rapidly in recent years.It can acquisition the 3D point cloud data of the measured object surface efficiently,quickly and accurately,multi-tense scanning can record the actual scene in different periods.In this paper,the data acquisition of multi-tense 3D laser scanning is carried out in the process of archaeological excavation.This paper studied Multi stage data registration,segmentation and feature extraction,and based on the principle of the real record of the relics in the archaeological disciplines,proposing "remains fidelity modeling,non-remains geometry modeling" modeling constraints,that is the fine modeling of the remains objects,the geometric modeling of the non-remains objects,discussed the three dimensional model construction method for archaeological site excavation scene.Finally,using the spatial analysis method based on data and model,this paper makes a summary of the cultural factors hidden in the representation of remains.Oriented to the point cloud data processing method for 3D reconstruction of the field archaeological excavation scene,this paper mainly completed the following research contents and results:(1)The basic principles and methods of multi-tense 3D laser point cloud data registration are discussed,the data acquisition and spatial registration of multi period point cloud data are realized by using FARO Focus3D laser scanner.(2)Proposed and implemented for the staged multi-phase point cloud data feature extraction:remains body,remains opening plane data extraction method and remains opening line extraction method;segmentation in a variety of object point cloud data segmentation in the test pit,and fill the lack of point cloud.(3)Under the "remains fidelity modeling,non-remains geometry modeling"modeling constraints,compared different modeling methods results,proposed combined use the TetGen method and least-square fitting plane method to build the fine model for test pit,moreover,discussion and realization the model connection,formation boundaries stitch method,complete the combination build of the test pit integrated model.(4)Based on the multi-tense point cloud data and the 3D model of the archaeological site scene,take Ling Jia Tan site archaeological excavation data as research object,with many types of archaeological knowledge comprehensive analysis the morphological characteristics of sintering soil 3D model and point clouds,reasoning and study on the behavior characteristics of ancient human dumping sintering soil,such as a dumping place,dumping mode etc.Space analysis of the scattered distribution of animal bones in the test pit,summarized the distribution characteristics and extension research on the related relics.