廣域全周俯瞰監視與視覺偵測; Wide-scoped Surrounding Top-view Monitoring and Visual Detection

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/61011

Please use this identifier to cite or link to this item: https://ir.lib.ncu.edu.tw/handle/987654321/61011

Title:	廣域全周俯瞰監視與視覺偵測;Wide-scoped Surrounding Top-view Monitoring and Visual Detection
Authors:	郭予中;Kuo,Yu-Chung
Contributors:	資訊工程學系
Keywords:	全周俯瞰監視;視覺偵測;智慧車輛;盲點;車輛安全;Wide-scoped Surrounding Top-view Monitoring;Visual Detection;Intelligent Vehicle;Blind-spot;Vehicle safety
Date:	2013-07-26
Issue Date:	2013-08-22 12:09:42 (UTC+8)
Publisher:	國立中央大學
Abstract:	近年來全周俯瞰監視系統已逐漸應用在減少車輛周遭盲點已降低碰撞事故的駕駛輔助。這些系統提供給駕駛在車輛周圍短距離的視野，而短距離的視野使得這些系統僅能夠使用在倒車及停車這類低速的應用。我們提出的廣域全周俯瞰監視與視覺偵測包含了一個建置廣域全周俯瞰影像的方法，能夠產生比現有系統更廣的視野；同時並對影像內容進行視覺偵測，偵測出畫面中的障礙物。此廣域全周俯瞰監視與視覺偵測系統包含：相機校正、多部相機影像對位、影像融合、己車移動向量估計、地面特徵移除及障礙物標示等模組。系統利用安裝於車輛前後以及兩側的相機，經過拍攝校正板對相機影像做校正並估計相機位置後，將影像投影在一個3D模型上，計算影像融合，產生一個連續無縫的全周影像。視覺偵測透過相機所的參數計算出車輛的移動向量，根據移動向量分辨出影像中的障礙物。我們在柏油路面、水泥路面和地磚路面上對系統進行測試。實驗結果顯示視覺偵測在水泥和地磚路面上能夠有效的移除非障礙物的地面特徵，而對比較低的柏油路面則容易造成車輛移動向量估算的錯誤而造成偵測錯誤。 In recent years, surrounding top-view monitoring systems are becoming a practical driving aid that help reducing collision hazards by eliminating blind spots. Many of such systems provide short range views surrounding the vehicle, limiting its application to parking and reversing. In this paper, we propose a practical system for creating a wide-scoped surrounding imagery around the vehicle and highlighting obstacles in the driving environment. By using a calibration board setup, the cameras we mounted on each side of the vehicle can be calibrated so the projections of the cameras form a continuous surround view on a dual-camber model. The projected imagery gives drivers the freedom to change view-point to suit different driving needs. By estimating the ego-motion of the vehicle using the input image sequence of the cameras, the proposed system is able to detect objects in the images by finding movements of features that do not correspond to ground motion relative to vehicle motion. Detected obstacles are highlighted in the wide-scoped surround top-view imagery to warn the driver of potential hazards. We tested our system against asphalt, concrete, and tiled road surfaces with obstacles in the scene. The results show while concrete and tiled surface features can be effectively removed, feature-poor asphalt surface is prone to misdetection for errors introduced during calibration and ego-motion estimation.
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	989	View/Open

社群 sharing

Loading...