Dynamic pedestrian detection method of lidar

Abstract

The invention discloses a dynamic pedestrian detection method of lidar in the technical field of road traffic safety. The method specifically includes the following steps: S1, lidar installation; S2,radar coordinate system setting; S3, lidar data clustering; S4, feature extraction; S5, feature classification; and S6, pedestrian sample collection. The method carries out clustering algorithm analysis for obstacles of laser detection, forms a feature set, has good precision and robustness for pedestrian detection, can detect a single person, can also accurately detect pedestrian targets for non-motorized vehicles and pedestrians moving side by side, and has the characteristics of a good detection effect and high detection precision.

Description

technical field [0001] The invention discloses a laser radar dynamic pedestrian detection method, in particular to the technical field of road traffic safety. Background technique [0002] As an important part of the intelligent transportation system, road traffic safety has been paid more and more attention by research institutions at home and abroad. In the mixed traffic environment, pedestrian detection is of great significance for safe assisted driving. In the current pedestrian detection methods, single-sensor methods are commonly used, or texture, entropy, template matching and other methods are used to detect moving pedestrians in general urban environments, or stereo vision systems are used to obtain ground objects of interest by using camera tilt compensation methods , and use support vector machines to detect pedestrians. However, the detection method based on a single sensor is limited by the small amount of information and the complexity of the algorithm. It is d...

AI Technical Summary

Problems solved by technology

In the current pedestrian detection methods, single-sensor methods are commonly used, or texture, entropy, template matching and other methods are used to detect moving pedestrians in general urban environments, or stereo vision systems are used to obtain ground objects of interest by using camera tilt compensation methods , and use the support vector machine to detect pedestrians. However, the detection method based on a single sensor is limited by the small amount of information and the complexity of the algorithm. It is difficult to meet the current rapid and robust pedestrian detection requirements.

Images