Pedestrian detection method

Abstract

The invention provides a pedestrian detection method. The pedestrian detection method comprises the following steps: acquiring a point cloud signal of a to-be-detected target by using a radar cross-sectional area feature clustering algorithm; carrying out denoising and clustering processing on the point cloud signal by utilizing a DBSCAN density clustering algorithm; generating an initial detection frame; obtaining a radar detection frame mapped by the initial detection frame in the visual coordinate system; training the target detection model by adopting a pedestrian database; generating a visual bounding box by using the target detection model; and performing data fusion on the radar detection frame and the visual boundary frame, and determining whether the to-be-detected target is a pedestrian target. According to the method, the point cloud signals are obtained through the intra-frame radar cross section (RCS) feature clustering algorithm, the point cloud signals are subjected to denoising and clustering fusion through the DBSCAN density clustering algorithm, denoising can be effectively and accurately achieved, effective signals in the point cloud signals can be reserved, the pedestrian target detection accuracy is improved, and the pedestrian target detection efficiency is improved. And meanwhile, the real-time requirement in the pedestrian detection task of automatic driving is also met.

Description

technical field [0001] The present application relates to the technical field of target detection, in particular to a pedestrian detection method. Background technique [0002] In recent years, autonomous driving has become one of the hotspots in the development of the automobile industry. It can not only effectively ensure traffic safety, but also provide driving services for people who do not have the ability to drive their own cars, generating huge social and economic value. More and more Many enterprises and scientific research institutions actively participate in and promote the development of autonomous driving. At the same time, considering the safety of pedestrians outside the vehicle, pedestrian detection technology has become an important research direction and a challenging research topic in autonomous driving. [0003] Pedestrian detection usually needs to find out all the pedestrians in the image or video frame, including location, distance information and size...

AI Technical Summary

Problems solved by technology

[0006] This application provides a pedestrian detection method, which can solve the problems that the current pedestrian detection technology is not accurate enough and the detection feedback response of pedestrian targets is not timely.

Images