Supercharge Your Innovation With Domain-Expert AI Agents!

Parallax acquisition method based on tree-shaped neural network structure, storage medium and computing device

A parallax acquisition and network structure technology, applied in the field of image processing, can solve the problems of difficult parallax calculation, poor processing effect of reflective areas and occluded areas, etc., and achieve the effect of solving occluded areas

Pending Publication Date: 2020-11-10
SOUTH CHINA UNIV OF TECH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] At present, most stereo matching algorithms are not effective in processing reflective areas and occluded areas. The reflective area will cause a big difference between the final parallax result and the actual one, and the pixels in the occluded area only appear in one image. , does not appear in another picture, so the disparity calculation is more difficult

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
  • Parallax acquisition method based on tree-shaped neural network structure, storage medium and computing device
  • Parallax acquisition method based on tree-shaped neural network structure, storage medium and computing device
  • Parallax acquisition method based on tree-shaped neural network structure, storage medium and computing device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0049] Such as Figure 1-Figure 4 As shown, a parallax acquisition method based on a tree-shaped neural network structure is applied to the left and right images acquired by a binocular vision system, including:

[0050] S1 Image preprocessing: Fourier transform the left image L and the right image R respectively, process the image features in the frequency domain, filter the image, and then perform inverse transformation, so that the noise of the image is removed and the image is reduced. Two images L of the effect of an overly bright reflective surface on the parallax result p with R p .

[0051] S2 extracts image features through a tree neural network: the processed left image L p with R on the right p The feature maps of the two images with different resolutions are extracted through the tree neural network respectively.

[0052]The tree-shaped neural network is similar to the binary tree in the data structure, except that each node of the last node has two child node...

Embodiment 2

[0081] This embodiment discloses a storage medium, which stores a program. When the program is executed by a processor, the method for acquiring parallax based on a tree neural network structure described in Embodiment 1 is implemented.

[0082] Parallax acquisition methods, specifically including:

[0083] S1 image preprocessing, specifically: perform Fourier transform on the left image L and right image R respectively, filter the image, and then perform inverse transformation to obtain two images L p with R p ;

[0084] S2 extracts image features through a tree-shaped neural network, left picture L p and right figure R p Eight feature maps F with different resolutions are obtained respectively 0 , F 1 , F 2 , F 3 , F 4 , F 5 , F 6 , F 7 ;

[0085] S3 takes the obtained eight different resolution feature maps as input through the hierarchical cost aggregation network, and finally obtains the refined final disparity map;

[0086] S4 For each pixel, select the disp...

Embodiment 3

[0089] This embodiment discloses a computing device, which includes a processor and a memory for storing a program executable by the processor. It is characterized in that, when the processor executes the program in the memory, the parallax acquisition method described in Embodiment 1 is implemented.

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 parallax acquisition method based on a tree-shaped neural network structure, a storage medium and computing equipment, and the method comprises the steps: denoising an imagethrough Fourier transform, extractingimage features of different resolutions through a tree-shaped deep neural network to acquire the parallax of the image; and obtaining a disparity map from the feature maps with different resolutions of the left and right maps through a hierarchical cost aggregation network. Compared with other methods, the invention can solve the influence of the reflective surface and the shielding area in the image on the parallax acquisition result.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a parallax acquisition method based on a tree-shaped neural network structure, a storage medium and a computing device. Background technique [0002] With the advancement of society's science and technology, the development of stereo matching technology is changing with each passing day. The effect and speed of obtaining disparity maps by traditional matching methods cannot meet the current requirements of high performance and accuracy. The development of deep learning has brought new solutions to stereo matching. Algorithms based on deep learning are well suited for stereo matching techniques. [0003] At present, most stereo matching algorithms are not effective in processing reflective areas and occluded areas. The reflective area will cause a big difference between the final parallax result and the actual one, and the pixels in the occluded area only appear in one image. , d...

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): G06T5/00G06N3/04G06N3/08G06T5/50
CPCG06T5/50G06N3/08G06T2207/10016G06T2207/20228G06N3/045G06T5/70Y02T10/40
Inventor 杜娟余政铭汤永超谭瀚儒孟靖雄
Owner SOUTH CHINA UNIV OF TECH
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com