A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Crack segmentation method and device based on Res-Unet multipath image feature multiplexing
The invention provides a crack segmentation method and device based on Res-Unet multipath image feature multiplexing, and the method comprises the steps: obtaining a to-be-processed pavement image, and carrying out the preprocessing of the to-be-processed pavement image, and obtaining a preprocessed pavement image; taking an image feature pyramid of the preprocessed pavement image; and extracting a first image feature of the preprocessed pavement image through a Res-Unet network obtained through pre-training, performing feature fusion on the first image feature and the image feature pyramid to obtain a target fused image feature, and processing the target fused image feature by using an attention mechanism to obtain a pavement crack segmentation result. According to the method, feature multiplexing can be better realized, so that the segmentation precision of the Res-Unet network is remarkably improved.
本发明提供了一种基于Res‑Unet多路图像特征复用的裂缝分割方法及装置，包括：获取待处理路面图像，并对待处理路面图像进行预处理得到预处理后路面图像；取预处理后路面图像的图像特征金字塔；通过预先训练得到的Res‑Unet网络，提取预处理后路面图像的第一图像特征，对第一图像特征和图像特征金字塔进行特征融合得到目标融合后图像特征，利用注意力机制对目标融合后图像特征进行处理得到路面裂缝分割结果。本发明可以更好地实现特征复用，从而显著提高Res‑Unet网络的分割精度。
Crack segmentation method and device based on Res-Unet multipath image feature multiplexing
The invention provides a crack segmentation method and device based on Res-Unet multipath image feature multiplexing, and the method comprises the steps: obtaining a to-be-processed pavement image, and carrying out the preprocessing of the to-be-processed pavement image, and obtaining a preprocessed pavement image; taking an image feature pyramid of the preprocessed pavement image; and extracting a first image feature of the preprocessed pavement image through a Res-Unet network obtained through pre-training, performing feature fusion on the first image feature and the image feature pyramid to obtain a target fused image feature, and processing the target fused image feature by using an attention mechanism to obtain a pavement crack segmentation result. According to the method, feature multiplexing can be better realized, so that the segmentation precision of the Res-Unet network is remarkably improved.
本发明提供了一种基于Res‑Unet多路图像特征复用的裂缝分割方法及装置，包括：获取待处理路面图像，并对待处理路面图像进行预处理得到预处理后路面图像；取预处理后路面图像的图像特征金字塔；通过预先训练得到的Res‑Unet网络，提取预处理后路面图像的第一图像特征，对第一图像特征和图像特征金字塔进行特征融合得到目标融合后图像特征，利用注意力机制对目标融合后图像特征进行处理得到路面裂缝分割结果。本发明可以更好地实现特征复用，从而显著提高Res‑Unet网络的分割精度。
Crack segmentation method and device based on Res-Unet multipath image feature multiplexing
基于Res-Unet多路图像特征复用的裂缝分割方法及装置
WANG SHIHAO (author)
2024-03-26
Patent
Electronic Resource
Chinese
IPC:
G06V
/
E01C
Bau von Straßen, Sportplätzen oder dgl., Decken dafür
,
CONSTRUCTION OF, OR SURFACES FOR, ROADS, SPORTS GROUNDS, OR THE LIKE
/
G06N
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
,
Rechnersysteme, basierend auf spezifischen Rechenmodellen
/
G06T
Bilddatenverarbeitung oder Bilddatenerzeugung allgemein
,
IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
MD-UNet: a medical image segmentation network based on mixed depthwise convolution
| Springer Verlag | 2024
Retinal vessel segmentation method based on RSP-SA Unet network
| Springer Verlag | 2024
UNet-based model for crack detection integrating visual explanations
| Elsevier | 2021