研究目的
To develop a technology for the automatic acquisition of the 3D information of road markings to meet the requirements of regular monitoring and maintenance tasks.
研究成果
The proposed cGAN-based P2P_L1 image translation method is effective and feasible for road marking extraction based on MMS point cloud data, and it exhibits advantages in terms of both accuracy and speed compared to the DeepLab V3+ advanced semantic segmentation algorithm.
研究不足
When the marking is damaged or partially covered by something (e.g. vehicles, pedestrians), false or missed extractions may occur.
1：Experimental Design and Method Selection:
The study uses ground roughness as a criterion to extract ground points based on the topological relationship of adjacent scan lines and generates feature images of a road surface using the adapted inverse distance weighted method. A finely adjusted image-to-image translation model named P2P_L1 is proposed for the segmentation of road markings.
2：Sample Selection and Data Sources:
Experimental data are collected by utilizing the SSW mobile mapping system developed by the Chinese Academy of Surveying and Mapping.
3：List of Experimental Equipment and Materials:
The SSW system is integrated with multiple sensors, including four SONY α7 cameras, a Riegl VUX-1HA laser scanner, an inertial measurement unit, and a global position system antenna.
4：Experimental Procedures and Operational Workflow:
The study involves ground filtering and feature image generation, image translation with P2P_L1 model, and 3D vectorization of road markings.
5：Data Analysis Methods:
The study compares the proposed model with the DeepLab V3+ network in terms of precision, F1-score, and mean Intersection over Union indicators.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
