Research And Implementation Of Feature Line Extraction Algorithm From Measured Point Cloud

In order to improve the work efficiency of archaeologists in making the scale drawing of cultural relics by hand,this paper proposes a feature line extraction algorithm for cultural relic point cloud,which can extract the closed feature line of cultural relic point cloud through the implementation of this algorithm.the closed feature lines drawn can be used as the scale drawing of cultural relics for follow-up research.At present,archaeological investigation,excavation and research must be completed by means of drawing,photography,measurement and other technical means,using drawings to reflect the size,shape,pattern and texture of cultural relics,so as to provide reliable data for scientific research.Therefore,archaeological mapping is an important and meticulous work.So far,the basic method of obtaining the first-hand drawing data of the archaeological site still needs to rely on the manual drawing of archaeologists.Based on this,this paper takes the cultural relic point cloud as the research object,and studies the method of effectively connecting the feature lines and drawing the closed structure diagram by using the valence information of the Euclidean minimum spanning tree(Euclidean Minimum Spanning Tree,EMST)nodes,so as to automatically draw the scale map of cultural relics from different perspectives.The main work is summarized as follows:(1)When extracting feature points,traditional methods usually use the derivative of principal curvature to determine the ridge and valley lines of objects,so as to extract features,but this method has high time complexity and poor anti-noise.In view of this,firstly,the KD tree algorithm is used to preprocess the point cloud source data to get the nearest neighborhood point set of the cultural object point cloud,then the curvature of the point set is estimated,and the Monge polynomial fitting method is used to find out the potential feature points.secondly,on the basis of the traditional noise reduction method,the Gaussian smoothing method is introduced to refine and reduce the noise of the potential feature points.The purpose of this paper is to do a good job of data preprocessing for the research of connecting characteristic lines.(2)In practice,it is found that the extracted point cloud contour feature points usually appear in the form of clusters,so this paper introduces the Euclidean minimum spanning tree algorithm to cluster the set of potential feature points extracted in the first step,by setting the threshold,retain the high-density potential feature point clusters,and remove the discrete feature points or low-density feature point clusters,which can also reduce the traversal time when connecting the feature lines.The smaller the threshold is,the more clusters are retained.Finally,the largest group of clustering results are taken as the point set connecting the feature lines.On the basis of the clustering results,by setting the connection conditions of the feature lines,the feature points in the point set are connected to the preliminary feature line of the cultural object point cloud.(3)In view of the fact that the extraction results of the existing point cloud feature line extraction algorithms can not be directly used as the scale map of cultural relics,this paper proposes a method of connecting the feature lines into a closed structure by assigning valence information to the nodes of the Euclidean minimum spanning tree.then,with the help of the node valence state information setting conditional design algorithm,the endpoint of each preliminary feature line is connected with the nearest feature point on the nearest feature line.So that it can form a complete structure map,in order to achieve the effect of automatically drawing the scale map of archaeological relics,and improve the work efficiency of archaeologists in drawing archaeological drawings.