An anti-collision warning method based on pedestrians and cyclists ahead of the road

Abstract

The invention discloses an anti-collision early warning method based on pedestrians and cyclists ahead of the road, which belongs to the field of driving assistance systems and includes three aspects of environment perception, information interpretation and target state judgment. The present invention sends the collected training set into the YOLO-R network for training, performs target detection and classification, and uses Kalman filtering to realize multi-target tracking; performs inverse perspective transformation on images to obtain IPM images, and simulates and synthesizes them through data regression The relationship curve between the original image and the pixel coordinates of the IPM image, and then the distance is estimated from the linear relationship between the pixel coordinates of the IPM image and the world coordinates; according to the speed of the vehicle and the braking distance, the early warning activation area is determined, and then the target in the activation area is used The fuzzy early warning algorithm judges whether the target is in danger, and if there is danger, it will remind the driver in time, so as to effectively reduce the occurrence of accidents and protect the safety of pedestrians and cyclists.

Description

Technical field [0001] The present invention belongs to the automotive active safety, involves knowledge of image processing and anti-collision warning systems, and specific to a non-collision warning method based on road front pedestrians and cysets. Background technique [0002] In this typical hybrid transportation environment, pedestrians, rides, vehicles, etc. are road traffic participants, including pedestrians, cavers are vulnerable groups, they exposed, no protective measures, personal safety during traffic accidents It is more difficult to guarantee, so the safety of the protection of pedestrians and rides is particularly important. [0003] Traditional pedestrian and cycling detection algorithms are artificial extracting HOG, HAAR, LUV and other features, training classifiers, and complete goals detection. This method can achieve better detection effects under certain conditions, but the characteristics of manual design are dim, the target attitude changes, and the dete...

AI Technical Summary

Problems solved by technology

[0004] Chinese patent (CN1O2765365A) discloses a pedestrian detection method based on machine vision and a pedestrian anti-collision warning system. A pedestrian classifier is used to detect pedestrians on the road, which blurs the individual characteristics of pedestrians and reduces the impact of individual differences on detection results. , judge the possibility of an accident through the anti-collision warning system, but its detection accuracy is low, ignoring the safety of the rider

Chinese patent (CN204870868U) discloses a car anti-collision and pedestrian protection early warning system based on multi-sensors, which uses laser ranging and ultrasonic ranging to realize early warning of vehicles, obstacles or pedestrians in front of and behind the car. The method has a single early warning indicator, The reliability of the early warning system is poor

Images