期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2005
卷号：XXXVI-3/W24
出版社：Copernicus Publications
摘要：Behavior analysis of each driver during congestion is still a problem in microscopic traffic analysis. To solve this problem, image sequences of traffic data are collected by helicopter. After detection and tracking of the vehicles, detailed data for microscopic traffic analysis result. To increase speed and reliability of vehicle detection and tracking, road detection is suggested. In this way, the search area for vehicle detection is limited to road boundaries while other moving objects such as vegetation or pedestrians are omit- ted. Road detection from aerial helicopter images is done by a line detection method based on edge detection and morphology opera- tions. Unwanted lines and other errors introduced during the line detection are removed by a threshold method. These thresholds are based on gray value and connected component labeling. After this improvement process, some parts of road lines are still missing, due to physical obstacles and the previous operations. With a Hough transform method line fragments are connected to each other in order to form elongated road lines. The image sequence data are taken by a Basler A101f digital camera installed on a helicopter. Results show that most road lines are correctly detected and connected but still some minor parts of the lines are not completely con- nected. Results can be improved by using a better technique for local maxima detection of the parameter space. The parameter space obvi- ously shows number and positions of line that are completely correct are calculated
关键词：road delineation; morphology; Hough transform; line detection; traffic data
