淡江大學機構典藏:Item 987654321/35025
English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 64178/96951 (66%)
造訪人次 : 9304644      線上人數 : 212
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library & TKU Library IR team.
搜尋範圍 查詢小技巧:
  • 您可在西文檢索詞彙前後加上"雙引號",以獲取較精準的檢索結果
  • 若欲以作者姓名搜尋,建議至進階搜尋限定作者欄位,可獲得較完整資料
  • 進階搜尋
    請使用永久網址來引用或連結此文件: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/35025


    題名: 使用主成分分析實作人臉影像搜尋系統
    其他題名: Developing an face retrieval system based on PCA
    作者: 黃柏鈞;Huang, Bo-jun
    貢獻者: 淡江大學資訊工程學系資訊網路與通訊碩士班
    葛煥昭;Keh, Huan-chao
    關鍵詞: 人臉特徵偵測;人臉特徵點擷取;人臉影像搜尋;Facial Feature Detection;Facial Feature Points Extraction;Face Retrieval
    日期: 2009
    上傳時間: 2010-01-11 05:55:25 (UTC+8)
    摘要: 近年來,人臉辨識技術已經被廣泛的利用在各方面並且取得不錯的成果。本論文提出一個完整的系統架構,根據使用者的搜尋選項例如整張人臉、眼睛、鼻子或是嘴巴來從資料庫中取出相似的人臉影像。在第一階段當中,系統擷取出人臉特徵,接著對其光源及大小做正規化,並且取出人臉特徵點。第二階段中,利用主成分分析以及人臉特徵相對關係從資料庫中搜尋出最相似的前三張人臉影像。實驗結果顯示出使用者可以根據其搜尋選項找到相似的影像,而人臉特徵相對關係可以彌補主成分分析容易受到光源影像的缺點。
    Over the last decade, face recognition techniques are widely used in various fields and have significant results. This paper proposes a system framework that retrieves the target image from the database according to the option of the users: whole face, eyes, nose or mouth. At the first stage, the system extracts the facial features, normalizes the scale and illumination of the feature images. From the feature images, we take advantage of the characteristic that facial features are conceived to be illumination invariant to filter the skin color then obtain the facial feature points. At the second stage, the system uses principle component analysis (PCA) and correlation of facial features to retrieve the target images from the face database. The result shows that users can find the similar face image according to their search option, and correlation of facial features can overcome the characteristic that PCA is sensitive to change in illumination.
    顯示於類別:[資訊工程學系暨研究所] 學位論文

    文件中的檔案:

    檔案 大小格式瀏覽次數
    0KbUnknown425檢視/開啟

    在機構典藏中所有的資料項目都受到原著作權保護.

    TAIR相關文章

    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library & TKU Library IR teams. Copyright ©   - 回饋