嵌入式車道偏移警示系統設計與實作;Design and Implementation of an Embedded Lane Departure Warning System

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/74680

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/74680

Title:	嵌入式車道偏移警示系統設計與實作;Design and Implementation of an Embedded Lane Departure Warning System
Authors:	吳祈寬;Wu, Chi-Kuan
Contributors:	資訊工程學系
Keywords:	車道偏移警示系統;車道偏移;車道偵測;決策樹;特徵分析;嵌入式;LDWS;Lane Departure Warning;Lane Detection;Decision Tree;feature analysis;Embedded
Date:	2017-07-24
Issue Date:	2017-10-27 14:36:02 (UTC+8)
Publisher:	國立中央大學
Abstract:	車道偏移警示系統是防範交通事故的主要駕駛輔助系統之一。本研究設計了一個嵌入式車道偏移警示系統，其作法包含影像前置處理、車道標記特徵分析、和左右車道線擷取、車道偏移判斷等四個主要功能模組。其中，車道標記特徵分析採用決策樹特徵分析，來篩選出候選車道標記，候選車道可進一步透過最小平方差進行直線參數偵測，以取得候選車道之直線模型，再以幾何特徵配對方式來找出最終車道線標記，最後再根據左右車道線來估計目前偏移量，在偏移量超過設定門檻值時，系統判定即將發生偏移，將發出警示提醒駕駛。對於真實環境的高雜訊道路影像，此一方法具有良好的強健性。本研究將系統實現於Raspberry Pi3嵌入式平台，以各種不同道路場景實際錄製之連續影像進行測試，實驗結果顯示出我們系統具有良好的可靠度，車道偵測正確率平均有84%以上，並且運作效能可達每秒30張即時需求。 ;The lane departure warning system is one of the main driving assistance systems to prevent traffic accidents. This study designed an embedded lane departure warning system, including image pre-processing, lane mark feature analysis, and left and right lane line extract, lane offset judgment and other four main functional modules. In this paper, the feature analysis of the lane mark is used to analyze the characteristics of the decision tree to select the candidate lane mark. The candidate lane can be further detected by the minimum squared difference to obtain the linear model of the candidate lane, and then the geometric characteristic pairing method The final lane mark, and finally according to the left and right lane line to estimate the current offset, the offset exceeds the set threshold, the system is about to be offset, the system will issue a warning to remind the driver. This method is robust in the real environment of high noise road image. In this study, the system is implemented on the Raspberry Pi3 embedded platform. The experimental results show that our system has good reliability, and the average correct rate of lane detection is more than 84% and operational efficiency of up to 30 real-time requirements per second.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	220	View/Open

社群 sharing

Loading...