| 3D reconstruction is a hot topic in the field of computer vision.It’s a theory and technology to study how to get the structural information of objects in three-dimensional space from two-dimensional images of objects.The current means of reconstruction are mainly divided into directly using the equipment to obtain three-dimensional information,using software to build three-dimensional model and 3D reconstruction based on image.The 3D reconstruction method based on images has the advantages of convenient data acquisition and low cost,which has been paid more and more attention in recent years.In addition,the array lens is widely used in computer vision and image processing because of its wider view and more useful information.3D printing technology does not require any mold.It can reproduce the three-dimensional effect of the object completely according to the 3D model data in the computer.It is also widely used in industrial production.In this paper,3D reconstruction techno logy based on images is deeply researched,and the 3D model reconstruction process based on human body image sequences is realized.Given a group of human array images,human body’s three-dimensional information can be obtained.The main steps are:collecting the image sequences of the human body,distortion correction and high dynamic processing,detecting and matching the feature points of the images,camera calibration and the 3D reconstruction of the human body.Then the reconstructed 3D model is derived to realize the 3D printing.The contents of this article are as follows:1.We set up an annular array lens device with a shooting angle of 360 degrees,control 36 shots,and photograph the characters in the scene at the same time,and get a set of array images of the human body.2.The SIFT operator with scale invariant features is used to detect and match the feature points between the human image sequences.3.The VisualSFM program is used to complete the self calibration of the array lens based on the motion recovery structure algorithm,and the internal and external parameters of the 36 lenses and the sparse point cloud of the human body are obtained.4.The CMVS/PMVS algorithm is used to restore the acquired sparse point cloud into a dense three-dimensional point cloud and process the dense point cloud texture by Meshlab software and save it into STL format.5.Using Repetier software,the dense point cloud data is converted into a gcode format after reconstruction process,and then a 3D printer is used to print out the 3D model of the human body.Based on the above methods,we demonstrate the 3D reconstruction process based on array lens and the printing implementation of 3D model based on the above methods,and give the experimental data and results of reconstruction and printing process.The 3D reconstruction and 3D printing based on this method are fast and effective,and have strong commercial application value. |
