Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Automated Framework to Audit Traffic Signs Using Remote Sensing Data
Traffic signs play a critical role in the safety and efficiency of any roadway; however, limited information exists on current traffic sign inventories (TSIs). The size of current traffic sign networks makes it economically challenging to apply traditional survey methods to the collection of a TSI. This paper proposes the use of light detection and ranging and video-log imaging to conduct a TSI, tested along a segment in Alberta, Canada. Signs along the segment were extracted through a Gaussian mixture model before measuring sign panel orientation. Then, the road surface near each traffic sign was extracted to measure its lateral and vertical placement. Next, sign classification was determined by applying a trained convolutional neural network. Finally, traffic sign visibility was measured to assess the time available for drivers to read and react to traffic signs. The extraction of traffic signs and the left and right lane markings had F1 scores of 95.1%, 94.4%, and 89.6%, respectively. The extractions were completed with a high degree of accuracy and with time benefits over traditional manual methods.
Automated Framework to Audit Traffic Signs Using Remote Sensing Data
Traffic signs play a critical role in the safety and efficiency of any roadway; however, limited information exists on current traffic sign inventories (TSIs). The size of current traffic sign networks makes it economically challenging to apply traditional survey methods to the collection of a TSI. This paper proposes the use of light detection and ranging and video-log imaging to conduct a TSI, tested along a segment in Alberta, Canada. Signs along the segment were extracted through a Gaussian mixture model before measuring sign panel orientation. Then, the road surface near each traffic sign was extracted to measure its lateral and vertical placement. Next, sign classification was determined by applying a trained convolutional neural network. Finally, traffic sign visibility was measured to assess the time available for drivers to read and react to traffic signs. The extraction of traffic signs and the left and right lane markings had F1 scores of 95.1%, 94.4%, and 89.6%, respectively. The extractions were completed with a high degree of accuracy and with time benefits over traditional manual methods.
Automated Framework to Audit Traffic Signs Using Remote Sensing Data
Karsten, Lloyd (Autor:in) / Gargoum, Suliman (Autor:in) / Saleh, Mohamed (Autor:in) / El-Basyouny, Karim (Autor:in)
29.04.2021
Aufsatz (Zeitschrift)
Elektronische Ressource
Unbekannt
Engineering Index Backfile | 1946
Engineering Index Backfile | 1933
Engineering Index Backfile | 1934
Engineering Index Backfile | 1933