Full metadata record
DC FieldValueLanguage
dc.contributor.author陳怡珊en_US
dc.contributor.authorI-Shan Chenen_US
dc.contributor.author林進燈en_US
dc.contributor.authorChin-Teng Linen_US
dc.date.accessioned2014-12-12T01:21:08Z-
dc.date.available2014-12-12T01:21:08Z-
dc.date.issued2007en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#GT009590512en_US
dc.identifier.urihttp://hdl.handle.net/11536/40108-
dc.description.abstract本論文提出一個快速而且實用性高的人臉識別系統,其用途在於偵測影像中的人臉位置並辨識該人臉是否為使用者。本系統包含人臉偵測(face detection),人臉識別(face identification)等。由於此系統以實作在即時系統上為前提來進行研究,及時處理(real-time)的要求極為嚴苛,計算量及精準度成為本論文的第一要求。系統的第一部份在偵測影像中人臉的位置,利用直方圖匹配的方法來對人臉資料庫做亮度上的正規化,接著使用2D haar 為特徵以AdaBoost 學習演算法並加以串聯式架構的觀念來進行人臉偵測器的訓練,同時提出以區域為基礎的分群法則來作為後端處理。而系統的第二部份在於辨識該人臉為使用者或入侵者,首先使用主成分分析方法,以找出重要的資訊及減少資料量的目的下來擷取人臉特徵,接著利用本論文提出的一個適應性機率模型(APM)來進行多個人臉上的識別,在APM 的設計上允許線上新增使用者以及同步更新使用者資訊。藉由以上提出的系 統,我們可以在複雜的環境下,進行多個人臉的偵測及識別,讓系統 在執行環境下更有彈性跟實用性。zh_TW
dc.description.abstractThis thesis tends to accomplish a fast and practical face recognition system. The target of this system is to locate face regions from a captured image and distinguish if these faces belong to any registered client. A multi-client system using adaptive probabilistic model (APM) is proposed to achieve a complete face recognition system which consists of the face detection unit and the face identification unit. In the face detection unit, an AdaBoost-based detector is implemented and improved by a lighting normalization technique and a region-based clustering method. In the face identification, the adaptive probabilistic model (APM) is proposed to establish a simple model for each registered client. Due to the design of APM, the proposed system can on-line add new clients and update the information of clients. Furthermore, the practicability and performance of the proposed system are demonstrated in the experimental results in this thesis.en_US
dc.language.isoen_USen_US
dc.subject人臉偵測zh_TW
dc.subject人臉辨識zh_TW
dc.subjectface detectionen_US
dc.subjectface identificationen_US
dc.subjectface recognitionen_US
dc.title使用適應性機率模型之多使用者辨識系統zh_TW
dc.titleMulti-client identification system using adaptive probabilistic model (APM)en_US
dc.typeThesisen_US
dc.contributor.department資訊學院資訊科技(IT)產業研發碩士專班zh_TW
Appears in Collections:Thesis


Files in This Item:

  1. 051201.pdf
  2. 051202.pdf

If it is a zip file, please download the file and unzip it, then open index.html in a browser to view the full text content.