AUTHOR=Gong Guoqiang , Liu Junqing , Li Zhengxiao 

TITLE=Vanishing Point Detection Method Based on Constrained Classification for Checkpoints on Urban Roads

JOURNAL=Frontiers in Bioengineering and Biotechnology

VOLUME=Volume 10 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/bioengineering-and-biotechnology/articles/10.3389/fbioe.2022.920329

DOI=10.3389/fbioe.2022.920329

ISSN=2296-4185

ABSTRACT=With the construction and development of smart cities, a great number of surveillance cameras have been widely used in the taking of traffic violation snapshots, vehicle tracking and other scenarios. The effective calibration of cameras installed at checkpoints on urban roads is the key to realizing the aforementioned applications, while the automatic detection of vanishing points is the main premise for realizing automatic camera calibration. The cascaded Hough transform (CHT), a classical method of detecting vanishing points, results in complicated and inaccurate calculation due to the large number of unorganized initial straight line data. To solve this problem, a vanishing point detection method based on constrained classification was proposed, which first filters short line data to avoid interference during straight line detection, before screening and classifying line segments via hierarchical clustering according to the image characteristics and the variation pattern of the angle similarity made by the straight line segments. Three types of straight line segments with the most significant angle differences were acquired. In order to prevent the optimization algorithm from getting stuck in the “wrong” local optimum neighborhood, or failing to locate the global optimum, a set of constraints were set to further restrict the search. Afterward, the classified line segments were projected into a finite rhombic space, which was then quantified. The point with the maximum vote was eventually identified as the vanishing point. The experimental results show that, compared with the classic method, not only has this method greatly reduced the computational complexity for vanishing points, but it also largely improved the accuracy of vanishing point detection.