본문 바로가기
  • Home

Development of Driving Lane Prediction Algorithm Utilizing Adaptive ROI based on Object

  • Journal of Software Assessment and Valuation
  • Abbr : JSAV
  • 2023, 19(3), pp.79-86
  • DOI : 10.29056/jsav.2023.09.09
  • Publisher : Korea Software Assessment and Valuation Society
  • Research Area : Engineering > Computer Science
  • Received : August 24, 2023
  • Accepted : September 20, 2023
  • Published : September 30, 2023

SungJu An 1 Chol Yong Soo 1

1신한대학교

Accredited

ABSTRACT

In this paper, we proposed a method of applying ROI to the lane prediction area in the lane recognition algorithm through the camera in autonomous driving. In the lane recognition method using a camera, lane recognition may be difficult due to environmental factors such as vehicles and obstacles on the road. In order to solve this problem, after getting an image through a camera, an ROI is allocated by estimating where lane detection is needed, and after removing the area of another vehicle adopting Deeplab-V3, lane recognition is performed through perspective, pre-processing, and Hough transformation. An algorithm that implements above transformations is proposed. A test image for the experiment used an urban driving image with many vehicle elements that can affect lane recognition. As a result of the experiment which setting the ROI for a specific area, it was possible to increase the accuracy of lane recognition by removing other vehicle areas and recognizing lanes.

Citation status

* References for papers published after 2023 are currently being built.

This paper was written with support from the National Research Foundation of Korea.