Method for reconstructing three-dimensional structured model based on any visual angle pictures

A technology of three-dimensional structure and pictures, applied in the field of computer vision

Pending Publication Date: 2021-07-06
SOUTH CHINA UNIV OF TECH
View PDF6 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The present invention uses a topology-adaptive graph convolution to solve the bottleneck of fixed topological relations in the deformation process in the prior art

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
  • Method for reconstructing three-dimensional structured model based on any visual angle pictures
  • Method for reconstructing three-dimensional structured model based on any visual angle pictures
  • Method for reconstructing three-dimensional structured model based on any visual angle pictures

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0070] The present invention will be further described below in conjunction with embodiment and accompanying drawing.

[0071] Such as figure 1 A method for reconstructing a three-dimensional structured model based on an arbitrary viewing angle picture is shown, comprising the following steps:

[0072] Step 1. Collect training data, including the 3D grid of the object and the corresponding RGB picture and depth picture.

[0073] Step 2. First segment the object parts. Different types of objects have different semantics, and then calculate the bounding box of each part. The parameters of the bounding box include: the coordinates of the center point, the length, width and height of the bounding box, and the length and width. Unit direction vector; and sample each segmented part, a total of 16890 points are taken according to the size of the area of ​​each triangular face. Such as figure 2 shown.

[0074] Step 3. Construct a depth map extraction network to extra...

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 discloses a three-dimensional structured model reconstruction method based on any view angle pictures. The method comprises the following steps of 1, collecting pictures, and forming a training data set; 2, segmenting object parts of the picture to obtain a bounding box of each part; 3, constructing a deep convolutional network model, and extracting a corresponding depth image from the image; 4, constructing a deep network model, and decoding the structural representation of the object; and 5, constructing and training a deep three-dimensional mesh deformation network, and deforming the bounding box into a three-dimensional model with a good structure. According to the method, the object structure representation is deformed, and a topology self-adaption method is used, so a problem that the topology is not changed when a general graph is used for convolution can be solved, and geometric details and the structure information of object recovery are ensured.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a method for reconstructing a three-dimensional structured model based on pictures of arbitrary perspectives. Background technique [0002] 3D reconstruction is a computer technique that uses projections of 2D information to restore 3D information. In the field of computer vision, 3D reconstruction has very high research value and is widely used in areas such as unmanned driving, artificial intelligence, SLAM, and virtual reality. Recently, 3D reconstruction methods based on deep convolutional neural network learning have gained popularity. Compared with traditional multi-view stereo algorithms, the learned model is able to encode rich prior information about the 3D shape space, which helps to resolve the ambiguity in the input. [0003] Although voxel-based and point cloud-based methods have been used for 3D reconstruction, these two representations cannot express the ...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06T17/20G06T19/20G06N3/04
CPCG06T17/20G06T19/20G06N3/044G06N3/045
Inventor 毛爱华代沧澜
Owner SOUTH CHINA UNIV OF TECH
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