Efficient Channel Attention Optimized Multi-layer Fusion Edge Detection Network for Boundary Extraction

DOI

10.2991/978-94-6463-516-4_47How to use a DOI?

Keywords

Edge Detection; Efficient Channel Attention; Multi-layer Fusion; Boundary Extraction

Abstract

In modern Building Information Modeling (BIM), precise boundary extraction is crucial for terrain construction and model generation. Addressing the limitations of current edge detection methods in boundary extraction from remote sensing images, this paper proposes a novel edge detection model named Efficient Channel Attention Optimized Multi-layer Fusion Edge Detection Network (EMF-NET). To retain more feature information during the network's downsampling process and improve the accuracy of boundary extraction, we integrate the Efficient Channel Attention (ECA) mechanism with max-pooling layers, creating the ECA Poolblock. The ECA Poolblock enables the network to more accurately identify target boundaries and structures during edge detection tasks, enhancing the precision and robustness of boundary extraction. Additionally, EMF-NET adopts a multi-layer end-to-end network architecture based on the concept of multi-value fusion, significantly outperforming traditional single-layer encoder-decoder architecture edge detection networks. Experimental results demonstrate that the proposed edge detection network achieves an F1 score of 90.18% and an Intersection over Union (IOU) of 80.78% in remote sensing image boundary extraction tasks on GF-2 dataset, markedly superior to other state-of-the-art edge detection methods, showcasing excellent edge detection performance.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Tianyi Qin
AU  - Bingyu Qu
AU  - Houlin Wang
PY  - 2024
DA  - 2024/09/17
TI  - Efficient Channel Attention Optimized Multi-layer Fusion Edge Detection Network for Boundary Extraction
BT  - Proceedings of the 2024 5th International Conference on Urban Construction and Management Engineering (ICUCME 2024)
PB  - Atlantis Press
SP  - 444
EP  - 454
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6463-516-4_47
DO  - 10.2991/978-94-6463-516-4_47
ID  - Qin2024
ER  -