Incremental multi-view three-dimensional reconstruction method

A 3D reconstruction and multi-view technology, applied in the field of computer vision, can solve problems such as the inability to incrementally complete 3D models, achieve the effects of simplifying the manual process and registration process, wide application range, and easy use

Inactive Publication Date: 2019-09-06
哈工大新材料智能装备技术研究院(招远)有限公司
View PDF0 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem of multi-view based on motion recovery structure, the present invention Figure three In order to solve the problem that the 3D model reconstruction process cannot use a single new image to incrementally complete the 3D model, an incremental multi-view based on feature matching and mapping rel

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Incremental multi-view three-dimensional reconstruction method
  • Incremental multi-view three-dimensional reconstruction method
  • Incremental multi-view three-dimensional reconstruction method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] an incremental multi-view Figure three Dimensional reconstruction methods, such as figure 1 As shown, the method includes:

[0041] Step 1. Multi-view based on motion restoration structure method Figure three 3D model reconstruction to obtain a 3D reconstruction model, and store the image path, image camera matrix and reconstructed 3D point information in the memory respectively;

[0042] Step 2, read the newly added image and the 3D reconstruction model described in step 1, establish the initial image pair according to the image feature matching relationship and the 2D and 3D mapping relationship, and reconstruct the 3D point cloud;

[0043] Step 3, adding an image to be reconstructed each time, and reconstructing the newly added image to be reconstructed until all images are reconstructed;

[0044] Step 4. Perform patch-based multi-view stereo vision dense reconstruction on the reconstructed sparse point cloud model, and then complete multi-view Figure three Dimens...

Embodiment 2

[0068] This embodiment demonstrates an incremental multi-view described in Embodiment 1 Figure three The specific reconstruction process of the three-dimensional reconstruction method in 12 multi-view images, figure 2 The 3D model after dense reconstruction from 12 multi-view images is displayed from three angles: left, front and right. It can be found that the structure on the right side of the wooden house has not been reconstructed. The specific process is as follows:

[0069] Execute step 1: store the image path, image camera matrix and reconstructed 3D point information in the multi-view reconstruction result into the memory respectively.

[0070] Execute step 2: read in the multi-view reconstruction results and 16 new images. Extract the SIFT features of the newly added image and the multi-view image, and perform feature matching between the newly added image and the multi-view image and between the newly added image groups. According to the number of matches and the ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention provides an incremental multi-view three-dimensional reconstruction method, belongs to the technical field of computer vision, and solves the problem of utilization of reconstructed model information in a multi-view three-dimensional reconstruction process. The method comprises the following steps: 1, obtaining a three-dimensional reconstruction model by utilizing a motion recovery structure method, and respectively storing an image path, an image camera matrix and reconstructed three-dimensional point information into a memory; 2, reading a new image and a three-dimensional reconstruction model, determining an initial image pair according to an image feature matching relation and a two-dimensional three-dimensional mapping relation, and reconstructing a three-dimensional point cloud; 3, newly adding one to-be-reconstructed image each time, and reconstructing the to-be-reconstructed newly added image until all the images are reconstructed; and 4, carrying out multi-view stereoscopic vision dense reconstruction based on a patch on the reconstructed sparse point cloud model to complete multi-view three-dimensional reconstruction. The method is high in applicability, andthe processes of manual image set screening and sorting and three-dimensional registration are simplified.

Description

technical field [0001] The invention relates to an incremental multi-view Figure three A three-dimensional reconstruction method belongs to the technical field of computer vision. Background technique [0002] With the development of electronic technology and computer technology, consumer-grade handheld camera equipment can easily obtain image information of objects from various angles. Multi-view based on motion recovery structure Figure three The three-dimensional reconstruction technology calculates and generates the three-dimensional model of the object by taking images of the object to be reconstructed from different angles. It has the characteristics of simple data acquisition, high degree of automation, strong adaptability, and wide application range. [0003] In the process of 3D reconstruction, this technology obtains the 3D model of the object to be reconstructed by inputting the multi-view images of all the objects to be reconstructed at one time. If the recons...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06T17/00G06T7/80G06T7/33G06K9/46G06K9/62
CPCG06T17/00G06T7/80G06T7/33G06V10/462G06F18/22
Inventor 金晶
Owner 哈工大新材料智能装备技术研究院(招远)有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap