摘要
通过观看交通监控视频可以有效获得有价值的交通信息参数,如行人及车辆运行的三维信息等。分析和处理这些交通参数并有效利用它们将对智能化交通和自动驾驶领域产生至关重要的影响。获得有效交通参数的一个非常重要的前提是必须对用于监控交通的相机进行标定,并利用相机的内外参数建立空间信息和图像信息的映射关系,从而获得有价值的交通信息参数。在当前交通场景下,基于消失点的相机自标定算法存在一些不足。例如,只有在交通场景中有斑马线、车道或建筑物的情况下,才能获得与其相关的消失点;容易受天气因素影响,导致检测被标定目标时精准度不高。基于上述问题,本文充分考虑到我们在城市道路场景中的行人及车辆运动目标以及在高速公路场景下的车辆运动目标来提取不同的消失点,并利用消失点相机自标定模型研究城市道路和高速公路场景标定问题。
By watching traffic monitoring video,valuable traffic information parameters,such as three-dimensional information of pedestrians and vehicles,can be effectively obtained.Analyzing and processing these traffic parameters and using them effectively will have a crucial impact on the field of intelligent mobility and autonomous driving.A very important prerequisite for obtaining effective traffic parameters is that the camera used to monitor traffic must be calibrated,and the mapping relationship of spatial information and image information must be established by using the internal and external parameters of the camera,so as to obtain valuable traffic information parameters.In the current traffic scenario,the camera self-calibration algorithm based on vanishing point has some shortcomings.For example,vanishing points associated with them can only be obtained if there are zebra crossings,lanes,or buildings in the traffic scene;It is easily affected by weather factors,resulting in low accuracy when detecting the target being calibrated;For when the vanishing point disappears and when the vanishing point of infinity or appears.Based on the above problems,this paper fully considers our pedestrian and vehicle movement targets in urban road scenes and vehicle motion targets in highway scenes to extract different vanishing points,and uses the vanishing point camera self-calibration model to study the calibration problems of urban roads and expressway scenes.
作者
田培辰
TIAN Peichen(Chang'an University College of Information Engineering,Xi'an Shaanxi 710000)
出处
《中国科技纵横》
2023年第2期15-17,共3页
China Science & Technology Overview
关键词
消失点
城市道路
高速公路
相机自标定
目标检测
vanishing point
urban roads
freeway
camera self-calibration
object detection
