Deep learning-based high beam identification method

A technology of deep learning and recognition method, applied in the field of high beams, can solve problems such as dazzling phenomenon, and achieve the effect of avoiding stray light interference, reducing feature redundancy, and reducing complexity

Pending Publication Date: 2021-03-12
合肥湛达智能科技有限公司
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] Aiming at the deficiencies of the prior art, the present invention discloses a high-beam recognition method based on deep learning, which is used to solve the problem that the driver's eyes are stimulated by the other party's high-beam and the dazzling phenomenon is likely to cause traffic accidents in the process of meeting cars at night. Provides a method of high beam recognition based on deep learning, which solves the shortcomings of traditional methods and significantly improves the recognition effect

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
  • Deep learning-based high beam identification method
  • Deep learning-based high beam identification method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0029] This embodiment discloses as figure 1 A high beam recognition method based on deep learning is shown, comprising the following steps:

[0030] S1 Set the camera in the area to be tested, and adjust its parameters such as height, pitch angle and camera exposure to meet the forensic distance;

[0031] S2 sets the appropriate exposure, highlights the halo area of ​​the car light, reduces the adhesion of the high beam in the image, and at the same time ensures the clarity of the vehicle license plate under the fill light;

[0032] S3 extracts features such as center brightness, edge symmetry, and linear gradient that characterize the image information of the light spot and halo, and judges the category of the car light;

[0033] S4 extracts and classifies car light categories based on multi-objective convolutional neural networks, and uses deep learning algorithms to achieve multi-task learning;

[0034] S5 uses the improved DBSCAN algorithm combined with the least square...

Embodiment 2

[0042] This embodiment discloses as figure 2 The vehicle high-beam recognition verification setting shown is a typical straight-going vehicle scene, H is the height of the monitoring pole, α is the camera’s overlooking angle, β is the vertical field of view of the lens, and points A, B, and C are the points on the road. mark point. Points C to A are the effective detection range of the high beam. When tracking the high beam vehicle between AC, when the high beam lasts for a long time or the driving distance is long, it is regarded as misusing the high beam.

[0043] In this embodiment, parameters such as camera height H, pitch angle α, and camera exposure are first set to meet the evidence collection distance. Set the appropriate exposure to highlight the halo area of ​​the car light, reduce the adhesion of the high beam in the image, and at the same time ensure the clarity of the vehicle license plate under the fill light. Secondly, the headlight area is preliminarily scre...

Embodiment 3

[0046] In this embodiment, when detecting night lights, extracting the multi-dimensional features of car lights requires separating the spot and halo area in advance. The average initial attenuation brightness of the high-beam car lights in the statistical database is used as the image segmentation threshold, and the narrow and long closed connected domains are filtered out. Preliminary detection The light spot of the car light, the light spot diverges outward in a circular shape to obtain the corresponding halo area. When the traffic flow is dense at night, the halos under the headlights and the left and right halos are easily damaged by ground reflections, body reflections, etc., and the linear structure jumps. Select the region of interest extracted from the upper half fan of the spot centroid as a linear feature, extract stable halo gradient features, and reduce stray light interference.

[0047] In this embodiment, when extracting car lights and visual object detection, t...

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 the technical field of high beams, in particular to a deep learning-based high beam identification method, which comprises the following steps of: extracting characteristics of center brightness, edge symmetry, linear gradient and the like for representing light spot and halo image information, and judging the type of a vehicle lamp; extracting and classifying vehicle lampcategories based on a multi-target convolutional neural network, and achieving multi-task learning by using a deep learning algorithm; combining an improved DBSCAN algorithm and a least square methodto carry out fitting on the vehicle lamp category; predicting an inter-frame displacement limit value to adaptively generate a tracking search area, and optimizing a target matching function of adjacent frames to achieve multi-vehicle lamp tracking; after the high beam lamp is tracked for a certain distance, performing left and right lamp pairing through colinear constraint, and completing high beam lamp identification. Stray light interference is avoided, and the feature redundancy is reduced. Meanwhile, the single vehicle lamp is independently matched up and down, so that the influence of vehicle lamp refitting is avoided, and the recognition effect is remarkably improved by using a deep learning algorithm.

Description

technical field [0001] The invention relates to the technical field of high beam lights, in particular to a method for recognizing high beam lights based on deep learning. Background technique [0002] The development of automobiles has injected vitality into China's economy, not only solving the employment problem, but also greatly facilitating people's lives. With the improvement of the economy, China's per capita income continues to grow, and more and more people have a higher pursuit of life. As a means of transportation, cars not only facilitate people's transportation, but also bring people a certain social "status". More and more people have begun to become "car owners". Statistics show that the number of cars per thousand people in my country is 114, while the United States, Germany, France and Japan have already exceeded 500, so there is still huge room for development in my country's auto market. [0003] While automobiles have greatly facilitated people's lives,...

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): G06K9/00G06K9/46G06K9/62G06N3/04B60Q1/14
CPCB60Q1/14G06V20/56G06V10/44G06N3/045G06F18/2321
Inventor 张中黄俊杰李安
Owner 合肥湛达智能科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap