Unlock instant, AI-driven research and patent intelligence for your innovation.

A free-view-angle three-dimensional object retrieval method and system

A retrieval system and object technology, applied in the field of computer vision, can solve the problems of limited application and the inability of deep neural networks to generalize application scenarios, and achieve the effect of enhancing feature vectors and stabilizing and reliable performance.

Active Publication Date: 2019-04-26
TSINGHUA UNIV +1
View PDF6 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] However, in the existing technology, the feature extraction method based on projection view needs to acquire images with a fixed camera angle, and then use the acquired image for feature extraction, which limits the feature extraction method based on projection view in the actual application process. Applications
The root cause of this problem is that when training a deep learning model for stereo object retrieval, for a fixed camera angle, there is a possibility of image feature vector overfitting, which makes the deep neural network unable to generalize to free angles to obtain video images. application scenarios

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
  • A free-view-angle three-dimensional object retrieval method and system
  • A free-view-angle three-dimensional object retrieval method and system
  • A free-view-angle three-dimensional object retrieval method and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0021] The following combination Figure 1 to Figure 2 The first embodiment of the present application will be described.

[0022] Such as figure 1 As shown, this embodiment provides a free-view stereo object retrieval method, including:

[0023] Step 1. Construct a retrieval neural network model based on the multi-view convolutional neural network and the projection view of the standard object;

[0024] This step 1 specifically includes:

[0025] Step 11. Acquire at least two projection views of different perspectives of at least one standard object in the image database;

[0026] Specifically, 12 projection views of different perspectives of m standard objects are acquired in the image database as the basic data for constructing the retrieval neural network model, where m is a positive integer greater than or equal to 1, and the specific value of m is determined by the model The required accuracy is determined.

[0027] Step 12: According to the multi-view convolutional neural network...

Embodiment 2

[0072] Such as image 3 As shown, this embodiment provides a free-view stereo object retrieval system 10, which includes: a network construction module 100, a database construction module 200, and a retrieval result output module 300;

[0073] The network construction module 100 is configured to construct a retrieval neural network model according to the multi-view convolutional neural network and the projection view of the standard object;

[0074] Further, the network construction module 100 specifically includes: an image acquisition unit 101, a feature extraction unit 102, a sorting unit 103, a zero-setting unit 104, and a model construction unit 105; the image acquisition unit 101 is used to acquire at least one standard in the image database At least two projection views of the object from different perspectives;

[0075] Specifically, 12 projection views of different perspectives of m standard objects are acquired in the image database as the basic data for constructing the re...

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 free-view three-dimensional object retrieval method and system, and the method comprises the steps: 1, building a retrieval neural network model according to a multi-view convolutional neural network and a projection view of a standard object; Step 2, extracting features corresponding to all standard objects in an object database according to the retrieval neural networkmodel, and constructing a retrieval feature database which comprises a plurality of retrieval three-dimensional objects; And step 3, according to the input Euclidean distance between the retrieval object and the retrieval three-dimensional object, outputting the retrieval three-dimensional object corresponding to the retrieval object. According to the technical scheme, the feature vectors extracted from the projection view retrieval network are enhanced, input of any position and any number of views can be supported for the projection view retrieval network, and stable and reliable performanceis achieved.

Description

Technical field [0001] This application relates to the technical field of computer vision, and in particular to a free-view stereo object retrieval method and a free-view stereo object retrieval system. Background technique [0002] With the advancement of technology, the collection and acquisition of three-dimensional object models has become easier and easier. At the same time, the data scale of three-dimensional objects is getting larger and larger. For example, Google 3D Warehouse, people's demand for three-dimensional object data retrieval is also increasing. The subjects have considerable potential value in many fields. Thanks to the tremendous progress made in deep learning and the rapid development of computer vision technology, the method of extracting features of stereo objects based on deep learning has achieved unprecedented development. These methods propose various methods for the various representation forms of stereo objects, such as voxels, point clouds and proj...

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): G06F16/583G06F16/53G06K9/62
CPCG06F18/22G06F18/2451G06F18/214
Inventor 高跃黄正跃马楠
Owner TSINGHUA UNIV