CN-121999637-A - Collision early warning method, device, equipment, storage medium and program product

CN121999637ACN 121999637 ACN121999637 ACN 121999637ACN-121999637-A

Abstract

The invention relates to the technical field of vehicle safety, and discloses a collision early warning method, a device, equipment, a storage medium and a program product, wherein first video frames acquired in different directions outside a vehicle are acquired; when determining that the movable objects exist in different directions of the vehicle at the same time according to the first video frame, acquiring real-time motion tracks of the movable objects in different directions; based on the real-time motion trail of each movable object, analyzing whether collision risks of trail intersection exist among the movable objects in different directions, if the collision risks exist, executing collision early warning actions on the movable objects, and judging whether the collision risks of trail intersection exist among the objects in different directions or not by synchronously acquiring and analyzing the real-time motion trail of the multi-directional movable objects around the vehicle, so that the technical problem that the real-time comprehensive judgment of multi-directional potential collision is difficult to be carried out from a third-party view in the related art is solved, and the active safety capability under a complex traffic scene is improved.

Inventors

SUN HAN
HUANG XUEYAN
TANG HUI
ZHANG YONGHAO
YAO ZAIQI
MA KAI
RONG HAI
TANG LI

Assignees

浙江吉利控股集团有限公司
吉利汽车研究院（宁波）有限公司

Dates

Publication Date: 20260508
Application Date: 20260407

Claims (10)

1. A collision warning method, the method comprising: Acquiring first video frames acquired in different directions outside a vehicle; When determining that the movable objects exist in different directions of the vehicle at the same time according to the first video frame, acquiring real-time motion tracks of the movable objects in different directions; Based on the real-time motion track of each movable object, analyzing whether collision risks of track intersection exist among the movable objects in different directions, and if the collision risks exist, executing collision early warning actions on the movable objects.
2. The method of claim 1, wherein the acquiring the first video frames acquired in different directions outside the vehicle comprises: the video frames in the front-rear direction of the vehicle are collected in real time by utilizing the front-rear side cameras of the vehicle configuration, and the video frames in the left-right direction of the vehicle are collected in real time by utilizing the left-right side cameras of the vehicle configuration; And determining the video frames in the front-rear direction and the video frames in the left-right direction of the vehicle as video data collected in different directions outside the vehicle, and determining a first video frame from the video data.
3. The method of claim 1, wherein the acquiring real-time motion trajectories of the movable object in different directions comprises: detecting the position coordinates of the movable object based on the first video frame, and inquiring whether an object identifier corresponding to the movable object is stored or not to obtain an inquiring result; and obtaining the real-time motion trail of the movable object according to the processing strategy corresponding to the query result.
4. The method of claim 3, wherein the obtaining the real-time motion trail of the moving object according to the processing policy corresponding to the query result comprises: if the query result is the stored object identifier corresponding to the movable object, acquiring a historical motion trail associated with the object identifier; matching the position coordinates with the historical motion trail; And if the matching is successful, generating a real-time motion track based on the position coordinates and the historical motion track.
5. The method according to claim 4, wherein the method further comprises: If the position coordinates are not successfully matched with the historical motion trail, accumulating the lost frame number of the movable object, and judging whether the lost frame number is larger than a loss threshold value or not; And if the lost frame number is smaller than or equal to the lost threshold value, matching the position coordinate with the historical motion trail again until the lost frame number is larger than the lost threshold value, deleting the object identification of the movable object and the historical motion trail, and re-executing the step of acquiring the real-time motion trail of the movable object.
6. The method of claim 3, wherein the obtaining the real-time motion trail of the moving object according to the processing policy corresponding to the query result comprises: If the query result is that the object identifier corresponding to the moving object is not stored, the position coordinate is used as a track starting coordinate, the object identifier is created for the moving object, and matching tracking is performed in a second video frame according to the track starting coordinate and the object identifier so as to generate a real-time motion track of the moving object, wherein the second video frame is a video frame with a time stamp later than that of the first video frame.
7. The method of claim 6, wherein the matching tracking in the second video frame according to the track start coordinates and the object identification to generate the real-time motion track of the moving object comprises: Generating a tracking target frame of the movable object in the second video frame, and taking the two-dimensional coordinate of the center point of the tracking target frame as the tracking position coordinate of the movable object; And associating the tracking position coordinates with the track starting coordinates to generate a real-time motion track of the movable object.
8. The method according to claim 1, wherein analyzing whether there is a collision risk of trajectory intersection between moving objects in different directions based on the real-time motion trajectory of each moving object comprises: Determining the appearance time length of the movable object based on the associated video frame after the first video frame; if the occurrence time is longer than the preset time, acquiring a time stamp and a corresponding abscissa of the movable object in each video frame according to the real-time motion trail; According to the time stamp and the abscissa, determining the time variation trend of the abscissa in a linear fitting mode; And determining the movement direction of the movable object based on the time variation trend of the abscissa, and analyzing whether collision risks of track intersection exist between the movable objects in different directions according to the movement direction.
9. The method of claim 8, wherein said determining the time trend of the abscissa by means of a linear fit comprises: using the time stamp in the motion trail as an independent variable and the abscissa as a dependent variable, and adopting a least square method to fit to obtain a fitting straight line; Constructing an error square sum function according to fitting parameters of the fitting straight line, wherein the fitting parameters comprise an abscissa parameter representing a starting position and an inclination parameter representing an inclination; and respectively carrying out deviation guide on the abscissa parameter and the inclination parameter to obtain an inclination value representing the time change trend of the abscissa.
10. A vehicle comprising a controller, the controller comprising a memory and a processor, the memory and the processor being communicatively coupled to each other, the memory having stored therein computer instructions, the processor executing the computer instructions to perform the method of any of claims 1-9.

Description

Collision early warning method, device, equipment, storage medium and program product Technical Field The invention relates to the technical field of vehicle safety, in particular to a collision early warning method, a device, equipment, a storage medium and a program product. Background The accident of the ghost probe is a typical and dangerous situation in road traffic, and refers to that a driver cannot timely find pedestrians or non-motor vehicles which suddenly cross a road due to the fact that the sight is blocked by obstacles such as adjacent vehicles and the like, so that the collision occurs due to insufficient reaction. Such accidents often result in serious personal injury. The related technical scheme is focused on monitoring based on dangerous vehicles or road-vehicle networking systems, the method has the defects of blind areas in monitoring vision, high system deployment and popularization cost and the like, and effective monitoring on the risk of a ghost probe accident from the perspective of a third-party vehicle is lacked, and early warning is realized. Disclosure of Invention The invention provides a collision early warning method, a device, equipment, a storage medium and a program product, which are used for solving the problem of how to effectively monitor the accident risk of a ghost probe from the perspective of a third party vehicle and realizing early warning. In a first aspect, the present invention provides a collision early warning method, the method including: Acquiring first video frames acquired in different directions outside a vehicle; When determining that the movable objects exist in different directions of the vehicle at the same time according to the first video frame, acquiring real-time motion tracks of the movable objects in different directions; Based on the real-time motion track of each movable object, analyzing whether collision risks of track intersection exist among the movable objects in different directions, and if the collision risks exist, executing collision early warning actions on the movable objects. The method comprises the steps of acquiring first video frames acquired in different directions outside a vehicle, acquiring real-time motion tracks of moving objects in different directions when the moving objects are determined to exist in different directions of the vehicle according to the first video frames, analyzing whether collision risks of track intersection exist among the moving objects in different directions or not based on the real-time motion tracks of each moving object, and executing collision early warning actions on the moving objects if the collision risks exist, wherein the collision risks of track intersection exist or not are judged by synchronously acquiring and analyzing the real-time motion tracks of the multi-directional moving objects around the vehicle. In an alternative embodiment, the acquiring the first video frames acquired in different directions outside the vehicle includes: the video frames in the front-rear direction of the vehicle are collected in real time by utilizing the front-rear side cameras of the vehicle configuration, and the video frames in the left-right direction of the vehicle are collected in real time by utilizing the left-right side cameras of the vehicle configuration; And determining the video frames in the front-rear direction and the video frames in the left-right direction of the vehicle as video data collected in different directions outside the vehicle, and determining a first video frame from the video data. According to the invention, video data in different directions are synchronously collected by utilizing the cameras at the front side, the rear side and the left side and the right side of the vehicle, the system can acquire real-time visual information around the vehicle, and the perception blind area of a single visual angle is eliminated. The fusion of the multi-source video data provides a complete data base for the follow-up accurate identification and tracking of moving objects (such as pedestrians and vehicles) in all directions, and greatly improves the overall perception capability of the system to complex traffic environments. In an optional implementation manner, the acquiring the real-time motion trajectories of the movable object in different directions includes: detecting the position coordinates of the movable object based on the first video frame, and inquiring whether an object identifier corresponding to the movable object is stored or not to obtain an inquiring result; and obtaining the real-time motion trail of the movable object according to the processing strategy corresponding to the query result. The method and the device can establish unique identification for a new object and start tracking by judging whether the detected movable object appears for the first time or is tracked, and can continue the historical track of the tracked object, so that each movable object