公开(公告)号：KR101383467B1

公开(公告)日：2014-04-08

申请号：KR1020130016017

申请日：2013-02-14

Applicant: 한국산업기술대학교산학협력단

Inventor： 김응태 ,  권오현

IPC: G06T7/00 ,  B60W30/12

Abstract: The present invention relates to a method for intelligent vehicles and, more particularly, to an adapting lane tracing method using vision based lane modeling production for extracting and tracing lanes in real time based on highly reliable lane models produced by using characteristics of a road video. The method comprises the following steps: a first step of inputting a video; a second step of setting a fixed region of interest (ROI) in the inputted video in the first step; a third step of transforming the fixed ROI set in the second step by the Hough transform method, extracting lanes and producing lane models based on extracted lanes; a forth step of setting a variable ROI by using the lane models produced in the third step; and a fifth step of outputting the variable ROI set in the forth step. Through the simple process, reliable lane models can be produced, and provided is the lane extracting and tracing method capable of processing in real time based on the lane models. [Reference numerals] (140) Scan operating unit; (AA) Hough transform lane extraction; (BB) lane model production; (S10) Inputting images; (S20) Setting a fixed ROI; (S40) Setting a variable ROI; (S50) Output

Abstract translation: 本发明涉及智能车辆的方法，更具体地，涉及一种使用基于视觉的车道建模生产的适应车道跟踪方法，其基于通过使用道路视频的特征产生的高度可靠的车道模型实时提取和跟踪车道。 该方法包括以下步骤：输入视频的第一步骤; 在第一步骤中在输入的视频中设置固定感兴趣区域（ROI）的第二步骤; 通过霍夫变换方法转换第二步中的固定ROI集的第三步骤，基于提取的车道提取车道并生成车道模型; 通过使用在第三步骤中产生的车道模型来设置可变ROI的第四步骤; 以及输出第四步中设定的变量ROI的第五步骤。 通过简单的过程，可以生成可靠的车道模型，并且提供了能够基于车道模型实时处理的车道提取和跟踪方法。 （附图标记）（140）扫描操作单元; （AA）霍夫变换车道提取; （BB）车道模型生产; （S10）输入图像; （S20）设定固定的ROI; （S40）设定变量ROI; （S50）输出