English  |  正體中文  |  简体中文  |  Items with full text/Total items : 889/889 (100%)
Visitors : 14785426      Online Users : 53
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Tips:
  • please add "double quotation mark" for query phrases to get precise results
  • please goto advance search for comprehansive author search
  • Adv. Search
    HomeLoginUploadHelpAboutAdminister Goto mobile version
    Please use this identifier to cite or link to this item: http://ccur.lib.ccu.edu.tw/handle/A095B0000Q/495

    Title: 基於PHF與MHOG混合特徵於行人檢測;Pedestrian Detection Based on PHF and MHOG Mixed Feature.
    Authors: 蘇及第;SU, JI-DI
    Contributors: 資訊管理系研究所
    Keywords: 行人檢測;特徵萃取;MHOG;SVM;部件遮蔽問題;pedestrian detection;feature extraction;MHOG;SVM;partial occlusion problem
    Date: 2017
    Issue Date: 2019-07-17
    Publisher: 資訊管理系研究所
    Abstract: 本研究提出一改進方法來解決單張影像行人檢測中,人潮擁擠或行人部分被遮蔽的情況。首先我們選用了兩種特徵萃取方法分別地使用於行人不同的部位,一是基於Gabor濾波前處理的MHOG特徵,此特徵萃取方法是延伸行人檢測的經典方法HOG,可有效增加此特徵對於行人上半身的描述;另一個是基於把原有的Haar-like特徵修改成適合行人部件使用的PHF特徵,因行走中的手臂與腿時常處於歪斜的狀態。兩種特徵萃取方法皆有使用積分圖加速法有效地加速運算。 得到行人特徵後,藉由兩層的SVM分類器來處理行人部分被遮蔽的情況,第一層SVM用來判斷各個部件是否被遮蔽,接著使用未遮蔽的部件所得到的機率分數作為對應的第二層SVM分類器的特徵值,即可判斷該檢測窗口是否有行人。最後使用ROC曲線及混淆矩陣兩個評估方式進行實驗,結果證明本研究提出之方法能有效處理人潮擁擠或行人部分被遮蔽的情況,並可以在不同的場景中成功地實現行人檢測。
    Pedestrian detection is a considerable practical interest. The study proposes an improved methodology to solve crowded scenes or partial occlusion problem in pedestrian detection. First, the study used two feature extraction methods in different parts of pedestrians. One feature descriptors is MHOG based on Gabor filter and the other is PHF. They modified from HOG and Haar-like features descriptors and are adaptive for the shape of human upper body and human limbs, respectively. Both feature extraction methods use the integral image to effectively speed up the step. Second, the two-layer SVM classifier is used to deal with the partial occlusion problem. The first layer SVM can be used to determine that which human part is occluded, and then let the probability scores that obtained by the unoccluded parts as the feature value of the second SVM classifier. Through this processing, we can determine whether the detection window contains pedestrians. The experiment was tested by the ROC curve and the confusion matrix and the experiment result demonstrate that the method for pedestrian detection can effectively solve crowded crowd or partial occlusion problem, and it can be achieved in different outdoor environments.
    Appears in Collections:[資訊管理系研究所] 學位論文

    Files in This Item:

    File Description SizeFormat

    All items in CCUR are protected by copyright, with all rights reserved.

    版權聲明 © 國立中正大學圖書館網頁內容著作權屬國立中正大學圖書館


    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback