English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 51776/87004 (60%)
造訪人次 : 8385557      線上人數 : 58
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: http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/111355


    題名: 利用LBP特徵於Ada-SVM進行表情辨識
    其他題名: Facial expression recognition using LBP on Ada-SVM
    作者: 徐正平;Syu, Jheng-Ping
    貢獻者: 淡江大學資訊工程學系碩士班
    顏淑惠;Yen, Shwu-Huey
    關鍵詞: 表情辨識;LBP特徵;Adaboost;SVM;Expression recognition;LBP feature
    日期: 2016
    上傳時間: 2017-08-24 23:50:44 (UTC+8)
    摘要: 表情辨識在電腦視覺中應用的範圍很廣泛,如嬰幼兒照護、老人照顧等人機互動的應用。有準確的表情辨識結果才能提高這些應用的效能。本篇論文提出以簡單的LBP (Local Binary Pattern) 為特徵,利用Adaboost (Adaptive Boosting) 和SVM (Support Vector Machine) 以提高辨識率。對一張人臉影像,正規化後切成大小不一的重疊區塊並取其 LBP長條圖作為紋理特徵。針對每一種表情,使用Adaboost挑選出具有辨識能力的特徵區塊,由這些特徵訓練SVM達到單一表情的辨識。我們使用兩種不同訓練方式以得到辨識結果: One-against-one SVM 和One-against-all SVM,在使用JAFFE以及CK兩個表情影像資料庫進行實驗,結果顯示,這兩種方式都可以得到不錯的辨識效果。
    Expression recognition in computer vision has a wide range of applications, such as use of infant care, care of the elderly and other human-computer interaction. Accurate facial expression recognition is a crucial key to the successfulness of these applications. In this paper, an algorithm using LBP (Local Binary Pattern) histograms together with Adaboost (Adaptive Boosting) and SVM (Support Vector Machine) is proposed. Given a normalized Image of human face and divide it into overlapping blocks of various sizes. The LBP histogram is used as a texture feature. For each expression, use Adaboost to select the discriminative blocks, and train SVM with these features to achieve recognition single expression. Then, a multiclass SVM is trained to yield a final recognition. We test the proposed algorithm on JAFFE and CK two face image databases and have promising experimental results.
    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 描述 大小格式瀏覽次數
    index.html0KbHTML28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