Responsive image
博碩士論文 etd-0903110-122827 詳細資訊
Title page for etd-0903110-122827
論文名稱
Title
以支持向量機為基礎之行人偵測辨識系統
Pedestrian Detection and Recognition System Using Support Vector Machines
系所名稱
Department
畢業學年期
Year, semester
語文別
Language
學位類別
Degree
頁數
Number of pages
87
研究生
Author
指導教授
Advisor
召集委員
Convenor
口試委員
Advisory Committee
口試日期
Date of Exam
2010-06-29
繳交日期
Date of Submission
2010-09-03
關鍵字
Keywords
支持向量機、行人偵測
Support Vector Machines, pedestrian detection
統計
Statistics
本論文已被瀏覽 5676 次,被下載 0
The thesis/dissertation has been browsed 5676 times, has been downloaded 0 times.
中文摘要
本文研究關於使用單一攝影機的動態行人偵測系統及靜態行人偵測系統。在靜態影像部分,本文重新建立樣本資料庫,特徵抽取部分採用HOG(Histogram of oriented gradient)結合SVM(Support Vector Machines)分類器,實驗結果顯示出本論文使用的樣本庫及演算法在不同場景中HOG 搭配SVM 皆能偵測到行人。 在動態影像部分,由於老年人口及弱勢族群逐漸增加,穿越路口對於他們是一項挑戰,因此本文研究使用單一攝影機的動態影像偵測系統以輔助自主式運輸機器人,偵測位於路口中的行人,協助老年人口及弱勢族群穿越斑馬線,其中動態影像偵測系統的演算法是使用腳步偵測。根據實驗結果,動態影像系統不論在縱向或橫向都具有滿足即時偵測的效率,但動態以及靜態實驗結果顯示仍然會受到光線以及穿著的影響。
Abstract
This study considers the dynamic pedestrian detection system and the static pedestrian detection system with a single camera. In the static detection system, this study reconstructs the static database. As to feature extraction, HOG combining with SVM classifier is used in this study. Experimental results show the database can detect people by this algorithm in several scenes. In the dynamic detection system, because the population of older persons and disabled persons increases gradually nowadays, cross the intersection is a challenge for older persons and disabled persons, so this study researches in dynamic pedestrian detection system by a single camera for assisting autonomous transport robots, and this system detects people at the intersection for assisting older persons and disabled persons when they cross the intersection. As to the algorithm this study uses the foot detection algorithm to detect dynamic pedestrians. According to the experimental results, the light and clothes effect on the experimental results both in the dynamic pedestrian system and the static pedestrian system. The dynamic pedestrian system still shows real-time performance not only in the longitudinal direction but also in the lateral direction.
目次 Table of Contents
誌謝 i
目錄 ii
圖目錄 v
表目錄 x
摘要 xi
Abstract xii
第一章 緒論 1
1-1 研究背景 1
1-2 文獻回顧 3
1-3 研究動機與目的 14
1-4 本文架構 14
第二章 行人偵測技術介紹 16
2-1灰階化 16
2-2 Sobel邊緣偵測法 18
2.3梯度角度統計圖 21
第三章 支持向量機介紹 25
3-1 簡介 25
3-2 線性可分支持向量機 26
3-3 線性不可分支持向量機 30
3-4 非線性可分支持向量機 33
第四章 系統軟硬體架構 38
4-1 系統硬體 38
4-1-1 Canon相機 39
4-1-2 影像擷取卡 41
4-1-3 可攜帶式DVD播放器 41
4-1-4 桌上型電腦 43
4-2 系統軟體 44
第五章 行人偵測 45
5-1靜態影像行人偵測 45
5-1-1影像輸入 47
5-1-2灰階化 47
5-1-3 Histogram of Oriented Gradients 47
5-1-4 SVM 51
5-2動態影像行人偵測 53
5-2-1影像輸入 53
5-2-2 Sobel 53
5-2-3腳步偵測 54
5-2-4 ROI選擇 55
5-2-5重複框取問題 56
第六章 實驗結果 58
6-1靜態影像實驗結果 58
6-1-1場景一:行人並排 58
6-1-2場景二:不同比例行人 59
6-1-3場景三:多人情況 61
6-2動態影像實驗結果 62
6-2-1動態橫向實驗場景一: 63
6-2-2動態橫向實驗場景二: 65
6-2-3動態縱向實驗場景: 67
第七章 結論與未來展望 69
7-1 結論 69
7-2 未來展望 71
參考文獻 72

參考文獻 References
參考文獻
[1]世界銀行http://www.worldbank.org/.
[2]高科技行車安全技術,取自;
http://itri.org.tw/chi/lib/DownloadFile.aspx?AttNBR=1310.
[3]淺談行人偵測(民96年3月)。機械工業雜誌。民99年6月29日,取自;http://vbie.eic.nctu.edu.tw/vol_13/tech2.htm.
[4]P. Viola, M. J. Jones, “Robust real-time face detection,” International Journal of Computer Vision, vol. 52, no. 2, pp. 137-154, 2004.
[5]S. Bota, S. Nedesvchi, “Multi-feature walking pedestrians detection for driving assistance systems,” IEEE Intelligent Transport Systems, 2008.
[6]S. J. Krotosky and M. M. Trivedi, “On color-, infrared-, and multimodal-stereo approaches to pedestrian detection,” IEEE Conference Intelligent Transport Systems, 2007.
[7]X. B. Cao, H. Qiao, J. Keane, “A low-cost pedestrian-detection system with a single optical camera”, IEEE Intelligent Transport Systems,vol. 9, no. 58 – 67, 2008.
[8]S. Munder and D. M. Gavrila. “An experimental study on pedestrian classification,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, no. 11, pp.1863-1868, 2006.
[9]S. Paisitkriangkrai , C. Shen, J. Zhang, “Performance evaluation of local features in human classification and detection,” IEEE Transactions on Pattern Analysis and Machine Intelligence, no. 4, pp. 236-246, 2008.
[10]林宗勳。Support Vector Machines簡介。民99年6月29日,取自:http://www.cmlab.csie.ntu.edu.tw/~cyy/learning/tutorials/SVM2.pdf.
[11] DTREG. June 29, 2010, from the World Wide Web: http://www.dtreg.com/svm.htm .
[12] 雷祖強,周天穎,萬絢,楊龍士,許晉嘉,”空間特徵分類器支援向量機之研究”,航測及遙測學刊,12卷2期,145-163頁,民國96年。
[13] Biano(n.d.). Support Vector Machine .June 29, 2010, from the World Wide Web:http://www.cmlab.csie.ntu.edu.tw/~cyy/learning/tutorials/SVM3.pdf.
[14]Wikipedia. June 29, 2010, from the World Wide Web:
http://en.wikipedia.org/wiki/Support_vector_machine.
[15]Chih-Wei Hsu, Chih-Chung Chang, and Chih-Jen Lin, December 20, 2007. " A Practical guide to support vector classification.” June 29, 2010, from the World Wide Web:
http://www.scribd.com/doc/1156609/Libsvm-Guide.
[16]Libsvm 使用教學。民99年6月29日,取自:
http://www.cmlab.csie.ntu.edu.tw/~cyy/learning/tutorials/libsvm.pdf.
[17] 林弘德,民92年4月18日,piaip的(lib)SVM簡易入門。民99年 6月29日,取自:http://www.csie.ntu.edu.tw/~piaip/svm/svm_tutorial.html.
[18]G. Ma, D. Muller, S. -B. Park, et al. ”Pedestrian detection using a singlemonochrome camera,” IET Intelligent Transport Systems, vol. 3, pp. 42-56, 2009.
電子全文 Fulltext
本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。
論文使用權限 Thesis access permission:校內校外均不公開 not available
開放時間 Available:
校內 Campus:永不公開 not available
校外 Off-campus:永不公開 not available

您的 IP(校外) 位址是 3.142.173.227
論文開放下載的時間是 校外不公開

Your IP address is 3.142.173.227
This thesis will be available to you on Indicate off-campus access is not available.

紙本論文 Printed copies
紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊,請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。
開放時間 available 已公開 available

QR Code