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

Dual-branch Outdoor Unstructured Terrain Segmentation Network Based on Parallax and Plane Fitting

An unstructured, terrain segmentation technology, applied in the field of computer vision, can solve problems such as incomplete application, achieve good terrain segmentation performance, and improve adaptability

Active Publication Date: 2021-05-07
DALIAN UNIV OF TECH
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

exist figure 1 It can be seen in (c) that there is some noise and discontinuity in the ground plane deviation map, so the results of direct threshold segmentation cannot be fully applied

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
  • Dual-branch Outdoor Unstructured Terrain Segmentation Network Based on Parallax and Plane Fitting
  • Dual-branch Outdoor Unstructured Terrain Segmentation Network Based on Parallax and Plane Fitting
  • Dual-branch Outdoor Unstructured Terrain Segmentation Network Based on Parallax and Plane Fitting

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0049] The specific implementation manners of the present invention will be further described below in conjunction with the accompanying drawings and technical solutions.

[0050] The invention performs ground fitting and ground deviation calculation based on Matlab, and then builds a network model based on the Caffe framework.

[0051] Step 1: Data preparation, such as figure 1 The disparity map shown in (b) and the corresponding feasible and obstacle region labels.

[0052] Step 2: Input a parallax image, perform ground plane fitting δ=αx+βy+γ according to known pixel coordinates x, y and parallax value δ, and obtain plane parameters α, β, γ.

[0053] Step 3: After estimating the above parameters α, β, γ, use this parameter to further calculate the estimation of the ground plane in the camera coordinate system in the parallax space:

[0054] δ p =αx+βy+γ

[0055] Among them, x, y are the pixel coordinates in the image coordinate system, and α, β, γ are the plane paramete...

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 belongs to the technical field of computer vision and provides a dual-branch outdoor unstructured terrain segmentation network based on parallax and plane fitting. In this method, ground fitting and calculation of ground deviation are performed firstly, and disparity maps and ground plane deviation maps with geometric characteristics are selected instead of color images as network input. The network architecture designed by this method is composed of two sub-networks: the terrain segmentation main network and the stability and adaptability enhancement module. The two modules complement each other, showing a good terrain segmentation effect, and have good robustness and environmental adaptability. .

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a dual-branch deep network for terrain segmentation in an unstructured outdoor environment, which uses a disparity map and a fitted ground plane deviation map as inputs. Background technique [0002] Currently, autonomous robot navigation in unstructured outdoor environments remains an open and challenging problem. Terrain segmentation is one of the core tasks of robot navigation, and it is the key for the robot to identify traversable areas and avoid obstacles. Unlike urban roads, the terrain in unstructured outdoor environments is complex, featuring a combination of multiple terrain types and obstacles. exist figure 1 In (a), the lighting conditions produce shadows and oversaturation. In addition, obstacles (trees, dry grass) have high visual similarity to dirt roads with fallen leaves, which pose a great challenge for terrain segmentation. [0003] In hu...

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 Patents(China)
IPC IPC(8): G06T7/10G06K9/00
CPCG06T7/10G06T2207/10012G06T2207/20081G06V20/38
Inventor 李豪杰张朋博王智慧
Owner DALIAN 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