Improvement of distant approaching vehicle detection based on perspective transformation using on-board rear view camera

Shiho Tanaka, Kenichi Yamada, Toshio Ito, Takenao Ohkawa

Research output: Contribution to conferencePaperpeer-review

1 Citation (Scopus)

Abstract

Recently, vehicles with a rear view camera have been increasing. The rear view camera can be utilized as a sensor monitoring collision from behind in driving scene. We have been developing the technology to prevent rear-end collision from backward by detecting approaching vehicles using the image of the rear view camera. In the conventional methods of vehicle detection, the camera-view images are often used. However, it is difficult to estimate the position of the distant vehicle with accuracy because the distance resolution of the camera-view image becomes lower as the object which projected to the image becomes far from the camera. In this paper, we propose an improvement method to detect the distant approaching vehicle with accuracy from the top-view image, which is projected to the road plane using perspective transformation. The object in the top-view image moves at the proportional displacement to its speed. Furthermore, the width of the object is constant regardless of the distance from a host vehicle. The proposed method can provide the accurate position of the distant vehicle using these features of the top-view image. We show the details of the proposed method and availability by the experiment using the actual images.

Original languageEnglish
Publication statusPublished - 2010
Externally publishedYes
Event17th World Congress on Intelligent Transport Systems, ITS 2010 - Busan, Korea, Republic of
Duration: 2010 Oct 252010 Oct 29

Other

Other17th World Congress on Intelligent Transport Systems, ITS 2010
Country/TerritoryKorea, Republic of
CityBusan
Period10/10/2510/10/29

ASJC Scopus subject areas

  • Automotive Engineering
  • Control and Systems Engineering
  • Transportation

Fingerprint

Dive into the research topics of 'Improvement of distant approaching vehicle detection based on perspective transformation using on-board rear view camera'. Together they form a unique fingerprint.

Cite this