Research On Simplification Method For Point Cloud Of Terracotta Warriors Based On Characteristic Line

The terracotta warriors are important carriers of Chinese excellent history and culture.With the improvement of optical perception and calculation ability,digital modeling based on 3D scanning is widely used in the protection and display of terracotta warriors.A large amount of data and redundant data reduce the efficiency of processing,transmission and display.The existing point cloud simplification methods often emphasize execution efficiency and neglect the maintenance of point cloud features.It is aimed at requirements for feature retention in the application and display of terracotta warriors point clouds in this article.A3 D point cloud simplification method based on deep learning is proposed.The feature points of 3D point cloud are extracted through the feature lines of 2D image.For non-feature points homogenized sampling is carried out.The related system is designed and developed.At the same simplification ratio,the feature preserving effect of point cloud of terracotta warriors is improved.The main research work includes:(1)This paper proposes a method for extracting feature points from 2D to 3D based on a deep neural network.Aiming at problems of loss of key points and tedious calculation overhead in point cloud simplification,a novel network based on VGG16 is proposed.It obtains the characteristic lines in 2D image and extracts the characteristic points of 3D point cloud according to the mapping relationship between 2D image and 3D point cloud.Through the method in this paper,the complexity of calculation is reduced and the feature points in the point cloud are accurately obtained.(2)This paper proposes an adaptive non-characteristic point simplification method.Aimed at problems of loss of geometric shapes and the existence of blank areas,it is proposed that a point cloud simplification method based on simplification ratio.When the simplification ratio is low,uniform sampling is adopted,and when the simplification ratio is high,farthest point sampling is adopted.A series of experiments show that this method not only avoids large-scale holes,but also maintains the proper geometric shape.(3)This paper designs a point cloud simplification prototype system based on characteristic lines.In response to the needs of storage and display of terracotta warriors point clouds,several functions such as point cloud data reading,feature point extraction and point cloud simplification have been realized.The experimental results show that the simplified system function design is reasonable,which can effectively achieve the requirements on feature line extraction and 3D point cloud simplification,so that reduce the scale of terracotta warriors data,ensure the surface geometric characteristics better,and improve the efficiency of storage,transmission and presentation of point cloud...