Open AccessSemi-automatic road lane marking detection based on point-cloud data for mapping
Author(s) -
Zhen Kang,
Qiao Zhang
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1453/1/012141
Subject(s) - point cloud , computer science , computer vision , artificial intelligence , point (geometry) , interface (matter) , filter (signal processing) , position (finance) , line (geometry) , rgb color model , convolution (computer science) , mathematics , artificial neural network , geometry , bubble , finance , maximum bubble pressure method , parallel computing , economics
Traditional manual method for mapping is time consuming and has low precision. In order to increase the efficiency, this paper proposes a semi-automatic road lane marking detection based on point-cloud data. This approach requires simple interface interaction, and since point cloud data is memory intensive, we convert it to RGB images using the intensity and height of the 3D point cloud. Then, in the interface, the user needs to click a point around the detection line and generate image regions of interest. Finally, the image is filtered using a specific filter to obtain a convolved value for each point, and the position with the largest convolution value is the point at the position of the reticle. In this way, we get a point on the road marking, and multiple point connections generate the road markings on the map. Multiple experiments were performed on data from different routes obtained by different sensors, such as veledyne-64, veledyne-16, to verify the effectiveness of road lane marking detection based on point cloud data for mapping..