Automatic classification and vectorization of road marking lines based on vehicle 3D point cloud

Abstract

The invention provides a method for automatic classification and vectorization of road marking lines based on a vehicle-mounted three-dimensional laser point cloud, comprising the following steps: using the collected vehicle-mounted trajectory information and cutting the vehicle-mounted laser point cloud into multiple point clouds according to a defined length and a defined width; KD tree, search the neighbor points of each point in the block point cloud through the KD tree and calculate its normal vector features; build a normal vector-based growth algorithm, divide the point cloud into pre-segmented volumes, and perform feature extraction on each pre-segmented volume , extract road pavement points by constructing the feature constraints of the pre-segmented volume; construct a growth algorithm based on point cloud intensity features to extract road marking lines from road ground points; construct a road marking line template, classify road marking lines according to the template and Vectorize. The automatic classification and vectorization method of road marking lines based on the vehicle three-dimensional laser point cloud of the present invention has a high degree of automation and good stability.

Description

technical field [0001] The invention belongs to the field of image recognition, and in particular relates to an automatic classification and vectorization method for road marking lines based on a vehicle-mounted three-dimensional point cloud. Background technique [0002] Autonomous driving and intelligent transportation systems play an important role in improving road traffic safety and alleviating urban traffic congestion. High-precision maps play a key role in automatic driving systems and intelligent transportation systems. For example, high-precision maps can improve automatic driving systems in harsh environments. Stability under weather conditions, road traffic marking lines are an indispensable part of high-precision maps, and the extraction and vectorization of road traffic marking lines are of great significance. The vehicle-mounted 3D laser scanning system is widely used in the 3D data collection of urban road scenes due to its good mobility and high data acquisit...

AI Technical Summary

Problems solved by technology

Homayounfar et al. (2019) combined a deep learning neural network with a probabilistic graphical model to realize the automatic recognition and vectorization of road edges. This network can only be used for simple road edge extraction

Images