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

Unmanned vehicle monocular vision positioning method based on image characteristic dimensionality reduction

A technology of image features and monocular vision, applied in image analysis, image data processing, graphics and image conversion, etc., can solve the problems of slow computing speed, large amount of data, poor robustness, etc., reduce the probability of mislocation and ensure robustness Sticky, robust effects

Inactive Publication Date: 2017-09-29
TONGJI UNIV
View PDF1 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the current monocular vision has poor robustness, large amount of data, and slow calculation speed when positioning the scene changes.

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
  • Unmanned vehicle monocular vision positioning method based on image characteristic dimensionality reduction

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0030] Such as figure 1 Shown, flow process of the present invention is:

[0031] 1. For the input image, the global feature description is performed through the deep convolutional neural network DCNN, and the third convolutional layer is extracted as the image feature.

[0032] For the input image, the global feature description is performed by a deep convolutional neural network DCNN. This method uses the AlexNet network in the Tensorflow framework. The network won the championship in the 2012 imagenet image classification competition. The network structure has 5 convolutional layers, and 3 fully connected layers. Each convolutional layer contains activation functions and local response normalization, and then After pooling. It has been proved by practice that the 64896-dimensional features extracted by the third-layer convolutional network (cov3) are the most robust, and the scene can still be recognized when the environment changes greatly.

[0033] 2. Select a part (a...

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 relates to an unmanned vehicle monocular vision positioning method based on image characteristic dimensionality reduction. The method comprises steps that 1), global characteristic description of all input images is carried out through a depth convolutional neural network DCNN, and a third convolutional layer is further extracted as an image characteristic vector; 2), a principal component analysis algorithm is employed, multiple times of training and test are employed, and dimensionality reduction of all the image characteristic vectors is carried out to acquire an image characteristic vector after dimensionality reduction and a repeatedly-usable dimensionality reduction model. Compared with the prior art, the method is advantaged in that sensor cost is low, an application scope is wide, and strong robustness, faster operation speed and reduced false-positioning probability are realized.

Description

technical field [0001] The invention relates to the field of unmanned vehicle positioning, in particular to a monocular vision positioning method for unmanned vehicles based on image feature dimensionality reduction. Background technique [0002] In a variety of environments, including scene changes caused by light changes, weather changes, object movement, season changes, etc., whether the robot can accurately locate is the basic and core issue for realizing the real intelligence of the robot. Commonly used sensors for robot positioning include: GPS, odometer, radar, and binocular cameras, but each has its own disadvantages. Monocular vision combined with deep learning has a wide range of applications, high positioning accuracy, strong anti-interference ability, and no cumbersomeness caused by binocular positioning. It is currently a popular positioning method. However, the current monocular vision has poor robustness, large amount of data, and slow operation speed when po...

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): G06T7/73G01C11/00G01C11/04G06T3/00
CPCG06T7/73G01C11/00G01C11/04G06T3/06
Inventor 陈启军张会刘明王香伟杜孝国
Owner TONGJI UNIV
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