Identification and Analysis of Potential Collision Risks of Vehicle-Pedestrian Interactions on Crosswalk using Computer Vision Technique

Byeongjoon Noh; Haechan Cho; Hyejin Lee; Dongho Ka; Hwasoo Yeo

doi:10.7470/jkst.2023.41.1.001

All Issue

2023 Vol.41, Issue 1 Next Page

Article

Identification and Analysis of Potential Collision Risks of Vehicle-Pedestrian Interactions on Crosswalk using Computer Vision Technique 컴퓨터 비전 기반의 횡단보도 내 차량-보행자 간 잠재적 충돌 위험 인식 방법 및 분석에 관한 연구

28 February 2023. pp. 1-18

PDF XML

Abstract

Most of the traffic accidents involving vulnerable pedestrians occur on crosswalks. To reduce the collision risk between vehicles and pedestrians, we need to identify the risky situations and find the factors affecting it. We propose a method identifying the potential collision risks of vehicle-pedestrian interactions near crosswalks by applying computer vision and artificial intelligence techniques on videos. First, we employ a mask R-CNN model and transformation matrix in OpenCV to recognize the precise positions of the road users (e.g. vehicle and pedestrian) in orthogonal view. Next, we track their trajectories using the modified Kalman filter method. After extracting their interactions from trajectories of road users, we analyze their potential collision risks near crosswalks. In our experiments, we collected the video footage from 9 cameras in Osan City, tracked road users in videos, and analyzed the potential collision risks between them. On this video, the accuracy of road user detection was about 90% for both the vehicles and pedestrians when the distance tolerance is 50cm. For the tracking accuracy, our modified Kalman filter outperformed the threshold and minimum distance method by 3%. In analysis part, we elicited the valuable information that the vehicle speed is high on a wide road without a speed camera and the pedestrian safety margin is small on unsignalized crosswalks. We further observed that around 50% of cars do not stop before crosswalk when pedestrians cross the un-signalized crosswalk without having special warning signs such as red urethane road pavement and school zone marking on road.

Keywords

behavioral feature estimation

computer vision

crosswalk

environmental factor

potential collision risk

보행자는 도로상에서 가장 위험에 자주 노출되는 도로 이용자 중 하나로써, 횡단보도와 같은 차량과 공간을 공유하는 영역에서 그 위험도가 높게 발생한다. 따라서, 횡단보도에서의 차량과 보행자의 충돌을 줄이기 위한 연구는 필요하다. 본 연구에서는 도로 위에 기설치된 카메라에서 수집된 영상 정보를 활용하여 횡단보도 영역 내 차량과 보행자의 잠재적 충돌 위험 상황을 인식하는 방법을 제시한다. 먼저, 영상 내 객체 탐지에서 위치 탐지에는 mask R-CNN 모델을 활용하였으며, 시점변환행렬을 활용하여 일반적으로 비스듬하게 획득되는 영상정보를 탑 뷰(top-view) 시점으로 변환한다. 다음으로, 객체의 추적(tracking)에는 칼만 필터를 기반으로 하는 추적방법론을 활용한다. 이처럼 획득된 객체 정보를 바탕으로 차량과 보행자의 행동 특징을 추정한 후, 이러한 정보를 분석하여 횡단보도 근처에서의 잠재적 위험 상황에 대한 분석을 수행한다. 본 연구의 실험에서는, 오산시 내 9개의 횡단보도에서 수집된 CCTV 영상을 활용하였으며, 실험 결과 도로 이용자의 위치 탐지는 위치 허용 오차가 50cm일 때 차량과 보행자 모두에서 평균 정확도가 약 90%였고, 변형된 칼만 필터를 이용하여 객체의 경로를 추적하는 방식은 기존에 사용하던 임곗값과 최소 거리 방식보다 평균 정확도를 3%만큼 향상하였다. 차량과 보행자 간 잠재적 충돌 위험 상황을 분석한 결과 차선 수가 많고 과속 단속 카메라가 없으면 그 횡단보도를 지나는 차량의 속력이 높고, 신호등이 없는 횡단보도에서 보행자 안전 간격이 낮은 것을 확인하였다. 신호등이 없는 횡단보도에서는 붉은색 우레탄의 도로포장 및 도로 위의 스쿨존 표식이 되어있지 않은 지역에서 50% 정도의 차량이 보행자가 횡단보도를 건너는 상황에서 정지하지 않는 것을 확인하였다.

키워드

행동 특징 추정

컴퓨터 비전

횡단보도

환경 요인

잠재적 충돌 위험

References

Akhter F., Khadivizand S., Siddiquei H. R., Alahi M. E. E., Mukhopadhyay S. (2019), Iot Enabled Intelligent Sensor Node for Smart City: Pedestrian Counting and Ambient Monitoring, Sensors, 19(15). 10.3390/s1915337431374837PMC6696396

Avinash C., Jiten S., Arkatkar S., Gaurang J., Manoranjan P. (2019), Evaluation of Pedestrian Safety Margin at Mid-block Crosswalks in India, Safety Science, 119, 188-198. 10.1016/j.ssci.2018.12.009

Boufous S., Murray C. J., Vos T., Lozano R., Naghavi M. (2016), 75 The Rising Global Burden of Road Injuries, Injury Prevention, 22(Suppl 2), A29.1-A29. 10.1136/injuryprev-2016-042156.75

Boukerch Haroun, Luo Qing Sheng, Li Hua Shi, Boukraa Sebti (2019), Vision Based People Tracking System, International Journal of Computer and Information Engineering, 13(11), 578-582.

Ferranti L., Brito B., Pool E., Zheng Y., Ensing R. M., Happee R. et al. (2019), SafeVRU: A Research Platform for the Interaction of Self-Driving Vehicles with Vulnerable Road Users, IEEE Intelligent Vehicles Symposium. 10.1109/IVS.2019.8813899

Gandhi T., Trivedi M. M. (2007), Pedestrian Protection Systems: Issues, Survey, and Challenges, IEEE Transactions on Intelligent Transportation Systems, 8(3), 413-430. 10.1109/TITS.2007.903444

Gitelman V., Balasha D., Carmel R., Hendel L., Pesahov F. (2012), Characterization of Pedestrian Accidents and an Examination of Infrastructure Measures to Improve Pedestrian Safety in Israel, Accident Analysis and Prevention, 44(1), 63-73. 10.1016/j.aap.2010.11.01722062338

Jiang X., Wang W., Bengler K., Guo W. (2015), Analyses of Pedestrian Behavior on Mid-block Unsignalized Crosswalk Comparing Chinese and German cases, Advances in Mechanical Engineering, 7(11). 10.1177/1687814015610468

Ke R., Lutin J., Spears J., Wang Y. (2017), A Cost-effective Framework for Automated Vehicle-pedestrian Near-miss Detection through Onboard Monocular Vision, Computer Vision and Pattern Recognition. 10.1109/CVPRW.2017.124

Kim B.-G., Jang I.-S., Lee G. (2011a), Real-time Road-Visibility Measurement Using CCTV Camera, J. Korean Soc. Transp., 29(4), Korean Society of Transportation, 125-138.

Kim S. H., Jang J.-A., Choi K. (2005), A Hierarchical Approach for Diagnose of Safety Performance and Factor Identification for Black Spots (Based on Suwon-city), J. Korean Soc. Transp., 23(1), Korean Society of Transportation, 9-20.

Kim S.-R., Bae Y.-G., Chung J.-H., Kim H. (2011b), Factor Analysis of Accident Types on Urban Street using Structural Equation Modeling(SEM), J. Korean Soc. Transp., 29(3), Korean Society of Transportation, 93-101.

Liu H., Dai J., Wang R., Zheng H., Zheng B. (2016), Combining Background Subtraction and Three-Frame Difference to Detect Moving Object from Underwater Video, OCEANS. 10.1109/OCEANSAP.2016.7485613

Lobjois R., Cavallo V. (2007), Age-related Differences in Street-crossing Decisions: The Effects of Vehicle Speed and Time Constraints on Gap Selection in an Estimation Task, Accident Analysis and Prevention, 39(5), 934-943. 10.1016/j.aap.2006.12.01317275774

Lytras M. D., Visvizi A. (2018), Who Uses Smart City Services and What to Make of It: Toward Interdisciplinary Smart Cities Research, Sustainability, 10(6). 10.3390/su10061998

Matsui Y., Hitosugi M., Takahashi K., Doi T. (2013), Situations of Car-to-Pedestrian Contact, Traffic Injury Prevention, 14(1), 73-77. 10.1080/15389588.2012.67851123259521

Noh B., Ka D., Lee D., Yeo H. (2021), Analysis of Vehicle-pedestrian Interactive Behaviors Near Unsignalized Crosswalk, Transportation Research Record, 2675(8), 494-505. 10.1177/0361198121999066

Noh B., No W., Lee D. (2018), Vision-based Overhead Front Point Recognition of Vehicles for Traffic Safety Analysis, UbiComp/ISWC. 10.1145/3267305.3274165

Noh B., No W., Lee J., Lee D. (2020), Vision-based Potential Pedestrian Risk Analysis on Unsignalized Crosswalk Using Data Mining Techniques, Applied Sciences, 10(3). 10.3390/app10031057

Oh C., Cho J., Kim J.-H., Oh J.-T. (2007a), Methodology for Evaluating Real-time Traffic Safety Performance based on Vehicle Trajectory Data Extracted from Video Image Tracking, The 56th Conference of KST, Korean Society of Transportation, 179-187.

Oh C., Cho J., Kim J.-H., Oh J.-T. (2007b), Methodology for Evaluating Real-time Rear-end Collision Risks based on Vehicle Trajectory Data Extracted from Video Image Tracking, J. Korean Soc. Transp., 25(5), Korean Society of Transportation, 173-182.

Onelcin P., Alver Y. (2017), The Crossing Speed and Safety Margin of Pedestrians at Signalized Intersections, Transportation Research Procedia, 22, 3-12. 10.1016/j.trpro.2017.03.002

Oxley J. A., Ihsen E., Fildes B. N., Charlton J. L., Day R. H. (2005), Crossing Roads Safely: An Experimental Study of Age Differences in Gap Selection by Pedestrians, Accident Analysis and Prevention, 37(5), 962-971. 10.1016/j.aap.2005.04.01715993827

Park J., Lee G., Oh C., Kim J. H., Yun D. (2021), Development of Evaluation Methodology for Rear Collision Situation Using Vehicle Sensor Data, J. Korean Soc. Transp., 39(6), Korean Society of Transportation, 826-837. 10.7470/jkst.2021.39.6.826

Sengar S. S., Mukhopadhyay S. (2017), Moving Object Detection Based on Frame Difference and W4, Signal, Image and Video Processing, 11(7), 1357-1364. 10.1007/s11760-017-1093-8

Sun X., Yao H., Zhang S. (2010), A Refined Particle Filter Method for Contour Tracking, Visual Communications and Image Processing 2010, 7744, 77441M. 10.1117/12.863450

Wu J., Xu H., Zhang Y., Sun R. (2020), An Improved Vehicle-pedestrian Near-crash Identification Method with a Roadside LiDAR Sensor, Journal of Safety Research, 73, 211-224. 10.1016/j.jsr.2020.03.00632563396

Wu J., Xu H., Zheng Y., Tian Z. (2018), A Novel Method of Vehicle-pedestrian Near-crash Identification with Roadside LiDAR Data, Accident Analysis and Prevention, 121, 238-249. 10.1016/j.aap.2018.09.00130265910

Yang Y., Ning M. (2015), Study on the Risk Ratio of Pedestrians’ Crossing at Unsignalized Crosswalk, COTA International Conference of Transportation Professionals. 10.1061/9780784479292.25726481748PMC4736878

Information

Publisher :Korean Society of Transportation
Publisher(Ko) :대한교통학회
Journal Title :Journal of Korean Society of Transportation
Journal Title(Ko) :대한교통학회지
Volume : 41
No :1
Pages :1-18
Received Date : 2022-04-25
Revised Date : 2022-05-30
Accepted Date : 2023-01-12
DOI :https://doi.org/10.7470/jkst.2023.41.1.001

Journal of Korean Society of Transportation ISSN:1229-1366(Print) 2234-4217(Online) 대한교통학회지

All Issue