Three-dimensional model classifying method based on feature matching

A three-dimensional model and classification method technology, applied in character and pattern recognition, special data processing applications, instruments, etc., can solve the problems of large workload of two-dimensional image classification, limited practical application scope, etc. Save time and effort, avoid the effect of multiple views

Active Publication Date: 2017-06-13
TIANJIN UNIV
View PDF3 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The main challenges currently facing the field of view-based 3D model classification are: most methods focus on similarit

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
  • Three-dimensional model classifying method based on feature matching
  • Three-dimensional model classifying method based on feature matching
  • Three-dimensional model classifying method based on feature matching

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0054] Example 2

[0055] The following describes the scheme in Embodiment 1 in detail in combination with specific calculation formulas and examples. For details, see the following description:

[0056] 201: Define the multi-view color view set of each object in the training data as the multi-view training model library SD (SetDatabase), such as figure 2 As shown, a view is randomly selected from the multi-view color view set of each object to obtain the initial single-view view, and the single-view set of all objects is defined as the single-view training model library VD (View Database), such as image 3 Shown

[0057] 202: In the multi-view training model library and the single-view training model library, extract the CNN features of the initial view set of each object to obtain the initial feature multi-view training vector set And category labels Initial feature single view training vector set And category labels

[0058] Among them, the CNN feature, also known as the convol...

Example Embodiment

[0110] Example 3

[0111] The following combined with specific experimental data, Figure 4 The feasibility verification of the schemes in Examples 1 and 2 is carried out, as detailed in the following description:

[0112] The database used in this experiment is the database ETH released by Taiwan University of China [9] . This is a real-world multi-view model database, containing 80 objects in 8 categories and 41 views for each object. In this experiment, 24 objects of 3 of each type are selected as the training set, and the remaining objects are used as the set to be classified.

[0113] Several parameters are involved in this experiment: number of iterations, weight coefficient λ 1 , Λ 2 And neighboring points k 1 , K 2 . In this experiment, the number of iterations is set to 10, and the weight coefficient λ 1 = 0.9, λ 2 =0.1 and the number of neighboring points k 1 = 2, k 2 = 5. Comparing the category label of the three-dimensional model with the original category label after ...

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 model classifying method based on feature matching. The method comprises the steps that Gaussian kernels of training vector sets are extracted and defined as multi-view-angle training Gaussian kernels and single-view-angle training Gaussian kernels, and the Gaussian kernels are defined as a multi-view-angle training feature library and a single-view-angle training feature library; a target function is constructed through multi-view-angle training features and single-view-angle training features and iterated to be minimum, and a feature matching matrix is obtained; a view is randomly drawn from a multi-view-angle color view set of all objects in to-be-classified data, initial single-view-angle views and class tags of all the objects are obtained, and after convolution neural network features of the single-view-angle views are extracted, the single-view-angle Gaussian kernels of the features are calculated and defined as the single-view-angle training feature library; the single-view-angle feature library is multiplied by a transfer function thereof, mapped features are obtained, the view angle training features are multiplied by the other transfer function thereof, mapped features are obtained, the cos distance between the features is calculated, and similarity is obtained. The limit that the features must be located in the same space is avoided.

Description

technical field [0001] The invention relates to the field of three-dimensional model classification, in particular to a three-dimensional model classification method based on feature matching. Background technique [0002] 3D models, as a more colorful multimedia data type than 2D pictures, are constantly improving and developing in recent years. On the one hand, the development of modeling tools, 3D scanners, and 3D graphics acceleration hardware has made it possible to access and generate high-quality 3D models. Especially the invention and use of Microsoft Kinect has effectively promoted and promoted this development trend. On the other hand, the development of computer graphics, the application of 3D models such as industrial product design, 3D scenes, virtual reality, etc. make 3D models widely disseminated and used. 3D Models in Entertainment [1] ,medicine [2] ,industry [3] The research and use of other application fields have been recognized. The ever-growing In...

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): G06K9/62G06F17/30
CPCG06F16/583G06F18/22G06F18/24G06F18/214
Inventor 刘安安师阳聂为之
Owner TIANJIN UNIV
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