Font Size: a A A

Using Structured Light Means For The Rapid 3D Reconstruction Of Non-rigid Objects

Posted on:2021-03-03Degree:MasterType:Thesis
Country:ChinaCandidate:H G ChangFull Text:PDF
GTID:2428330623465018Subject:Computer technology
Abstract/Summary:PDF Full Text Request
Vision-based 3D reconstruction is an important research topic in computer vision domain,it can be divided into passive vision or active vision method according to whether light signal is emitted or not.The structured light technology is an active vision method with the advantages of non-contact,high precision,high speed and low cost,which makes it a research focus in 3D reconstruction domain and widely used in industry.Since the traditional 3D reconstruction methods mainly focus on the research of static or rigid objects,there are still many problems in the 3D reconstruction of non-rigid objects.The work of this paper is mainly aimed at the rapid 3D reconstruction technology of non-rigid objects based on structured light,and the research is carried out around the 3D clothing digital technology and real-time 3D reconstruction technology.The main work can be divided into the following three aspects:(1)Research a 3D clothing digital technology based on the RealSense depth cameras and develop the corresponding system.In this paper,three depth cameras and a turntable are used to build a 3D clothing scanning system,which can obtain point clouds from multiple views of clothing.A calibration method of camera pose based on SIFT algorithm and ICP algorithm is proposed.Based on this method and the calibration method of turntable,the pose parameters between cameras and turntable can be obtained.By using these parameters for the registration of point clouds of clothing,a complete clothing point cloud can be obtained.In this paper,a 3D clothing digital system is developed,and a variety of styles of clothing are reconstructed.The experimental results show that the system can quickly obtain a 3D clothing model with high precision under the condition of less depth cameras.(2)Research a real-time 3D reconstruction technology based on a near-infrared high frequency binary coding structured light algorithm.In this paper,an industrial camera and a near-infrared DLP projector are used to construct a structured light scanning system.Based on the existing binary coded structured light algorithm,this paper achieves the purpose of real-time 3D reconstruction by using the projector to project the high frequency coding pattern and executing the reconstruction calculation process in GPU.In this paper,a real-time 3D reconstruction system is developed to scan and reconstruct non-rigid objects in real time and compare with a variety of depth cameras.The experimental results show that in the indoor scene,the reconstruction speed of the system can reach 29 fps and the reconstruction accuracy can reach 0.05 mm,which is obviously better than the existing methods.(3)Research a multi-view real-time 3D scanning and reconstruction technology based on real-time 3d reconstruction technology.In this paper,three structured light scanning systems are used to construct a multi-view 3D scanning system.This paper solves the problem of mutual interference between the projected patterns of different systems by making each structured light system project and receive different wavelengths of light respectively.In this paper,a calibration method of camera pose based on a checkerboard corner detection technique is proposed for real-time registration of point clouds from different views.A multi-view real-time 3D reconstruction system is developed for non-rigid objects.Relevant experiments and error analysis are carried out for the above three works respectively,and visual results are obtained,which prove the feasibility and reliability of each system.In addition,there are some limitations in this paper,such as the texture color difference correction in the 3D clothing digital system,and the reconstruction speed of the real-time 3D reconstruction system.
Keywords/Search Tags:structured light, non-rigid body, real-time, 3D reconstruction
PDF Full Text Request
Related items