CN

Xiaolong Luo

Associate Professor       Supervisor of Master's Candidates

  • E-Mail:
  • School/Department:地球科学学院
  • Education Level:With Certificate of Graduation for Doctorate Study
  • Gender:Male
  • Degree:博士
  • Professional Title:Associate Professor
  • Discipline:Cartography and Geographical Information System
  • Enrollment disciplines: Cartography and Geographic Information Engineering Cartography and Geographical Information System , Computer Application Technology , Computer Science and Technology , Software Engineering ,

Paper Publications

Current position: Home > Scientific Research > Paper Publications

嵌入注意力机制的车道线像素级识别算法研究

Release time:2025-03-25
Hits:
Impact Factor:
1.078
DOI number:
10.14016/j.cnki.jgzz.2025.02.106.
Affiliation of Author(s):
长江大学
Teaching and Research Group:
GIS
Journal:
激光杂志
Key Words:
注意力机制;深度神经网络;语义分割;车道线识别;图像分割
Abstract:
车辆自动行驶的安全性和稳定性离不开车道线准确识别。然而,日常驾驶中面临着复杂多变的天气和光照条件、道路标记模糊或遮挡等挑战。研究并设计基于深度神经网络的车道线识别算法,以提高识别技术在面对复杂环境的鲁棒性与检测结果精度。通过构建以VGG-16 为主链并嵌入通道注意力和空间注意力机制的全卷积神经网络模型,实现端到端像 素级别的车道线语义分割。嵌入注意力模块的新模型在CULane 通用数据集上验证结果同VGG-解码语义分割方法相比,其平均像素准确率与均交并比(Mean Intersection over Union,MIoU)分别提升2.2%与1.3%。且在车道线不存在场景下,预测结果的像素准确率达到70%。 嵌入注意力机制的图像分割算法研究为车道线识别问题提供了有效解决方案,有力支撑车道线检测技术在无人驾驶场景的应用。
Indexed by:
Journal paper
Discipline:
Engineering
Volume:
46
Issue:
2
Page Number:
106-114
ISSN No.:
0253-2743
CN No.:
50-1085/TN
Date of Publication:
2025-02-01