Omnibearing pedestrian collision early warning method based on fisheye lens

Abstract

The invention provides an omnibearing pedestrian collision early warning method based on fisheye lenses, which comprises the following steps: a, installing a plurality of fisheye lenses with specifiedangles around a vehicle, and the viewing angle ranges of adjacent fisheye lenses are at least partially overlapped; b, collecting pedestrian images around the vehicle in real time through the multiple fisheye lenses; c, calculating human body semantic segmentation on the images acquired in real time by using a trained classifier to obtain positions of segmentation regions of different parts of the same human body under an image coordinate system; d, converting the position information obtained in the step c from an image coordinate system to a world coordinate system through a projection transformation matrix to obtain a corresponding position of a target human body or a target human body part on an XY plane; e, calculating an Euclidean distance equal value between the target contour coordinate and / or the 2D frame coordinate and the central point of the vehicle in a world coordinate system in a tracking period; and f, evaluating the early warning grade according to the data calculatedin the step e.

Description

technical field [0001] The invention relates to the field of machine vision, in particular to an omnidirectional pedestrian collision warning method based on a fisheye lens. Background technique [0002] Pedestrian detection is an important link in the field of safe driving, because the uncertainty of pedestrian movement is extremely high, different heights, short, fat and thin, clothing fabrics of various patterns, various postures of various age groups may appear, walking speed can change at any time, and the trajectory It is also not as clear as other types of goals. Driving in crowded conditions requires more careful handling. One-third of the annual traffic accident deaths in China are pedestrians. In order to solve this problem, high-end assisted driving systems or automatic driving systems need to be able to pre-warn or directly intervene in dangerous situations and wrong human driving behaviors, corresponding to the function of frontal pedestrian collision warning. ...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the above-mentioned defects of the prior art, the present invention proposes an all-round pedestrian collision warning method based on a fisheye lens. The fisheye camera installed around the vehicle can solve the problem of monitoring the surrounding road conditions of the whole vehicle. The human body semantic segmentation method of the network can well solve the problem that the pedestrian images overlap each other and only some human bodies appear, etc., which are difficult to identify. On this basis, the more accurate pedestrian distance can be obtained to make the early warning function of the system more comprehensive and accurate.

Images