視訊分析應用於智慧型運輸系統中之先進車輛控制與安全服務; Video Analysis for Advanced Vehicle Control and Safety Services in Intelligent Transportation Systems

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/9150

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/9150

Title:	視訊分析應用於智慧型運輸系統中之先進車輛控制與安全服務;Video Analysis for Advanced Vehicle Control and Safety Services in Intelligent Transportation Systems
Authors:	鄭旭詠;HSU-YUNG CHENG
Contributors:	資訊工程研究所
Keywords:	智慧型運輸系統;車道偵測;環境分類;先進車輛控制與安全服務;Advanced Vehicle Control and Safety Services;Lane Detection;Intelligent Transportation Systems;Environment Classification
Date:	2006-11-27
Issue Date:	2009-09-22 11:42:04 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	先進車輛控制與安全系統是智慧型運輸系統中很重要的一環。在視訊處理技術的協助之下，以圖像為基礎之車輛控制與安全系統可分析所拍攝之視訊並自動找出前方之車道與車輛，進而警告駕駛人車到偏離或是即將發生碰撞之危險情況。唯有正確地找出車道邊緣之所在位置，系統才能判斷目前車輛是否偏離車道中心，或是離同一車道之前方車輛距離過於接近。因此車道偵測是車輛控制與安全系統中重要之關鍵技術。本篇論文提出了一個階層式的車道偵測方法，以用於處理不同類型的道路。因為不同類型的道路往往需要不同的車道偵測方法，因此，我們先將道路環境分成兩大類: 結構道路與非結構道路。結構道路與非結構道路之分類係利用特徵值解構區別分析。分類完畢之後，不同類型的道路即用不同的方法來處理。本篇論文所提出之方法可有效地區分結構道路與非結構道路並針對不同類型的道路使用不同的處理方法。同時針對行車流量大之結構道路，設計了一套演算法可以有效地找出結構道路之車道的左右邊界，且不會被路面上其他往來的車輛所干擾影響。同時，實驗結果也顯示此方法於各種不同之照明情況下皆可以適用。本篇論文的主要貢獻包括了以下幾點: 首先，我們提出了階層式車道偵測的概念，可以充分利用針對各種路況所設計的方法來處理不同的道路環境。同時，利用特徵值解構區別分析，我們可以在訓練樣本有限的情況之下求得所需的高思函數並得到很好的分類結果。另外，在車道線偵測的部份，我們所設計的方法不會受光線及佔據路面車輛的影響。最後，對車流量較大的結構道路，我們所設計的方法能夠有效去除往來車輛的影響。 Advanced Vehicle Control and Safety Systems (AVCSS) play an important role in the Intelligent Transportation Systems. Lane boundaries have to be determined accurately in order to decide weather the vehicle is deviating from its current lane, or the vehicle is too close to the vehicle in the same lane in front of it. Therefore, lane detection is a crucial part in the advanced vehicle control and safety systems. This dissertation proposed a hierarchical lane detection system that can handle various types of road conditions. Because methods suitable for different road types are often different, we first classify the environments into two groups: structured roads and unstructured roads. Structured roads and unstructured roads are classified based on Eigenvalue Decomposition Regularized Discriminant Analysis (EDRDA). After classification, different lane detection algorithms are applied to different types of roads. In this dissertation, we are able to distinguish different road types effectively and use different algorithms to handle them. For structured roads, we propose a mechanism which is able to robustly find the left and right boundary lines of the lane and would not be affected by the passing traffic. Experimental results also show that the proposed method can work well in various lighting conditions. The main contribution of this dissertation includes several aspects. First, we propose the concept of hierarchical lane detection which can deal with different environment with relevant methods. Second, we apply EDRDA and design a voting mechanism for environment classification. With limited number of training samples, we are able to estimate the parameters for the Gaussian models using EDRDA and obtain satisfying classification results. Third, for structured roads those have heavier traffic, we design a mechanism to effectively eliminate the influence of passing vehicles when performing lane detection. We also extract lane-mark colors in a way that is not affected by illumination changes and the proportion of space that vehicles on the road occupy.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...