(3.236.214.19) 您好!臺灣時間:2021/05/10 08:21
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果

詳目顯示:::

我願授權國圖
: 
twitterline
研究生:紀欣呈
研究生(外文):HSIN-CHENG CHI
論文名稱:梅花型金字塔取樣演算法應用於人臉辨識
論文名稱(外文):An efficient algorithm for face recognition based on quincunx pyramid
指導教授:王敬文
指導教授(外文):Jing-Wein Wang
學位類別:碩士
校院名稱:國立高雄應用科技大學
系所名稱:光電與通訊研究所
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2009
畢業學年度:97
語文別:中文
論文頁數:81
中文關鍵詞:適應性奇異值分解光線補償法梅花型金字塔取樣法對數極座標取樣法
外文關鍵詞:Adaptive singular value decomposition (ASVD)quincunx pyramid samplinglog-polar mapping
相關次數:
  • 被引用被引用:3
  • 點閱點閱:315
  • 評分評分:系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
在這篇論文裡,我們提出一個新穎的架構,用以降低人臉辨識時所受到的光線變化及人臉姿態改變之影響。首先,我們提出適應性奇異值分解光線補償法,以RGB彩色空間個別分佈之平均值為依據,藉以調整光線補償係數的權重。其次,傳統對數極座標取樣法極為依賴取樣中心點,因而較無法排除人臉姿態變化之影響,而我們設計的梅花型金字塔取樣法則可有效克服此問題。經由FERET資料庫100位人臉之實驗,結果顯示我們提出的架構,其正確接受率高達99.8%,而錯誤接受率則為0%。
Face recognition often suffers from the lighting variation problem when dealing with face data with pose variations. Lighting compensation for the uneven illuminations on human faces can efficiently solve this problem. Different color spaces possess different characteristics and have been applied for by different recognition tasks. Recent research efforts reveal that color may provide useful information for recognition. In this thesis, we propose a novel methodology, adaptive singular value decomposition (ASVD)-based lighting compensation, to handle this problem. Following by the largest component of RGB color space, we accordingly multiply the other two components by adaptive weights based on their ratios among means.
The log-polar image geometry was first motivated by its resemblance with the structure of the retina of some biological vision systems and by its data compression qualities. In the last years, it has been noticed that the log-polar geometry also provides important algorithmic benefits to face recognition. However, it remains a problem how to locate the center point precisely on the human nose. In this thesis, we further propose applying an innovative quincunx pyramid sampling method with invariance to scaling, translation, and rotation to overcome this problem. PCA (Principal Component Analysis) and self-PCA, respectively, are subsequently applied to the two principal subspaces for discriminant analysis. With the discriminant information selected by genetic algorithms, our method can take full advantage of useful discriminant information in the face space.
Experiments using the color FERET (The Face Recognition Technology) database show the effectiveness of the proposed ASVD lighting compensation and quincunx sampling method. In particular, for the conducted experiment of 100 persons, which contains 200 training images, 100 query images, and 700 intruder images, the proposed method achieves the face recognition rate of 99.8% at the false acceptance rate of 0%.
摘 要 2
ABSTRACT 4
致謝 6
目 錄 7
圖索引 9
表索引 10
第一章 緒論 11
1.1 研究背景 11
1.2 研究動機及目的 11
1.3 相關文獻探討 13
1.4 章節概要 14
第二章 人臉辨識系統 17
2.1 簡介 17
2.2 人臉偵測與T型人臉擷取 17
2.3 適應性奇異值分解光線補償 19
2.4 人臉影像取樣 22
2.4.1 對數極座標取樣 22
2.4.2 梅花型金字塔取樣 25
2.4.3 大小、平移、和旋轉不變 27
2.4.4 熵(entropy) 28
2.5 特徵挑選與辨識 29
第三章 實驗結果與討論 49
3.1. 測試資料庫與實驗環境介紹 49
3.2. 實驗資料庫成員類別 50
3.3. 結果與討論 51
3.4. 其它文獻比較 52
第四章 結論與未來展望 65
4.1. 結論 65
4.2. 未來展望 65
參考文獻 66
附錄一 68
[1]R. Zhao, A. Chellappa, P. Rosenfeld, and P. Phillips, “Face recognition: a literature survey,” ACM Computing Surveys, pp. 399-458, 2003.
[2]H. Demirel and G. Anbarjafari “Pose Invariant Face Recognition Using Probability Distribution Function in Different Color Channels ,”Signal Processing Letters, IEEE, Volume 15, pp 537 – 540, 2008.
[3]L. H. Koh, S. Ranganath, and Y. V. Venkatesh, “An integrated automatic face detection and recognition system,” Pattern Recognition, vol. 35, no. 6, pp. 1259-1273, Jun. 2002.
[4]J. W. Wang, “Framework combined face segmentation with recognition,” Optical Engineering, vol. 48, no. 4, pp. 047206-1~047206-15, Apr. 2009.
[5]http://face.nist.gov/colorferet/faq.html
[6]詹明達,「假人臉排除與超解析技術用於人臉偵測與辨識」國立高雄應用科技大學光電與通訊研究所碩士論文,2008.
[7]張辰宇,「使用融合鑑別器於人臉辨識系統」,國立高雄應用科技大學光電與通訊研究所碩士論文,2006。
[8]Z. Sun, X. Yuan, G. Bebis, and S. J. Louis “Neural-network-based gender classification using genetic eigen-feature extraction,” IEEE International Joint Conference on Neural Networks, 2002.
[9]Srdjan and Miodrag “Characterization of Visually Similar Diffuse Diseases form B-Scan Liver Images Using Nonseparable Wavelet Transform,” IEEE Transaction no Medical Image, Volume 17, No.4, 1998.
[10]J. W. Wang, “Efficient facial component extraction for detection and recognition,” The 18th International Conference on Pattern Recognition (ICPR2006), Hong Kong, pp.20-24, 2006.
[11]C.-P. Liao and J. T. Chien, “Maximum confidence hidden Markov modeling for face recognition,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 30, no.4, pp. 606-615, April 2008.
[12]http://www.opencv.org.cn
[13]http://opencv.willowgarage.com/wiki/CvReference
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
系統版面圖檔 系統版面圖檔