Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Binocular parallax calculation method based on convolutional neural network

A convolutional neural network and binocular disparity technology, applied in the field of binocular disparity calculation based on convolutional neural network, can solve problems such as inability to obtain accurate disparity, lack of feature points, etc., to improve calculation accuracy, improve receptive field, Applicable effect

Active Publication Date: 2020-07-28
SOUTH CHINA UNIV OF TECH
View PDF2 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The current deep learning network models mainly include end-to-end convolutional neural network models and convolutional neural networks combined with traditional stereo matching algorithms. These network models are prone to lack of feature points when calculating parallax points in low-texture and reflective areas. without getting accurate parallax

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
  • Binocular parallax calculation method based on convolutional neural network
  • Binocular parallax calculation method based on convolutional neural network
  • Binocular parallax calculation method based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0033] Such as figure 2 and image 3 As shown, a binocular disparity calculation method based on convolutional neural network includes the following steps:

[0034] S1 uses the expansion cascaded convolutional network module to extract image features and obtain the left image feature data F L and the feature data F in the right figure R :

[0035] The expansion cascade convolutional network module has a three-layer structure, the first layer is a 3*3 convolution kernel layer, the second layer is three 1*1 convolution kernel layers and three 3*3 expansion convolution kernels The layers are combined in parallel, and the third layer is a 3*3 convolution kernel layer. This expanded cascaded convolutional network can effectively extract the disparity information of different receptive field ranges, and perform feature fusion, which can improve the richness of image feature information.

[0036] In the second layer, a 1*1 convolution kernel layer is the first parallel channel, ...

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 binocular parallax calculation method based on a convolutional neural network. Feature extraction is carried out on left and right images by using parallel small expansion convolution kernels; meanwhile, during multi-scale feature fusion, image edge features extracted by a Prewitt operator are added to reinforce the edge feature information of the binocular image; and then 4Dcost volume is constructed by using multi-scale feature information in combination with a parallax network layer, and finally cost aggregation is performed by using a 3D CNN module to obtain a parallax result of the binocular image.

Description

technical field [0001] The invention relates to the field of parallax of stereoscopic image pairs, in particular to a binocular parallax calculation method based on a convolutional neural network. Background technique [0002] The existing disparity acquisition methods for stereo image pairs mainly include traditional stereo matching methods and deep learning network training network model methods. The traditional stereo matching method has a certain difficulty in obtaining real-time stereo disparity due to the large amount of calculation and time-consuming. The disparity calculation method of the deep learning network, through a large amount of data training in the early stage, can quickly and accurately obtain the stereo disparity of the binocular image pair when the binocular disparity is obtained in actual use. The current deep learning network models mainly include end-to-end convolutional neural network models and convolutional neural networks combined with traditiona...

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): G06T7/593G06T7/564G06K9/62G06K9/46
CPCG06T7/593G06T7/564G06T2207/10012G06T2207/20081G06T2207/20084G06V10/454G06F18/253Y02T10/40
Inventor 杜娟汤永超谭笑宇
Owner SOUTH CHINA UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com