應用小波轉換與支持向量機之即時臉部識別系統

An interactive system that recognizes face on real-time using 2D wavelet transform and SVM

鄧氏金姣(Kim-Giao DangThi)

指導教授 ： 賀嘉生

中原大學/工學院/資訊工程研究所/碩士(2012年)

https://doi.org/10.6840/cycu201200073

摘要

在這篇論文中介紹了一個基於視訊的即時的人臉識別系統。所開發的系統可在受測對象使用配備有網路攝影機的筆記型電腦時識別該對象。該系統可用以增進智慧型人機互動（HCI），一旦使用者已被識別，人機互動系統可以客製化相應的偏好設定，授予或拒絕存取受控制的資源。為了讓我們的系統成功地識別人臉，本系統必須處理一些難題：諸如由於光照，姿態，與表情的變化而造成的面部外觀的連續的不受控制的變化。為了實現這一目標，本系統首先使用Viola-Jones演算法來偵測人臉以得到適當的紀錄。已紀錄的面孔，依照明，大小等因素進行正規化。接著，以離散小波變換應用於已正規化的面孔上，來找到該面孔的特徵。最後，將特徵輸入到多重類別支持向量機分類器以識別出具有這些特徵的人臉。我們已進行了10個受測者的封閉資料集合的實驗。實驗結果表明，該系統能夠達到較高的識別率。此外，它能夠以即時的方式執行面部特徵和人臉的偵測。

關鍵字

人臉辨識 ； 支持向量機 ； Viola-Jones 演算法 ； 離散小波轉換

並列摘要

In this thesis, I present a real-time video-based face recognition system. The developed system
identifies subjects while they are using a laptop with webcam. This system can be used to
improve smart human-computer interactions (HCI): once the user has been identified, an HCI
system can customizes the preferences accordingly, or grant or denies access to the controlled
resources. To allow for successful recognition, our system has to handle challenges such as
continuous, uncontrolled variations of facial appearance due to illumination, pose, and
expression. In order to achieve this, the system first detects the face for proper registration, using
Viola-Jones algorithm. The registered faces are then normalized in term of illumination, size, etc.
After that, discrete wavelet transform is applied on normalized faces to find the features
representing for the faces. Finally, the features are inputted to a multi-class SVM classifier to
identify the face represented by the feature. We have conducted closed-set experiments on a
database of 10 subjects. The experimental results show that the proposed system is able to reach
high recognition rate. Besides, it is able to perform facial feature and face detection, and
recognition in real-time.

並列關鍵字

SVM ； discrete wavelet transform ； Viola-Jones algorithm ； face recognition

參考文獻

AL-Jawad, N. (2009). Exploiting Statistical Properties of Wavelet Coefficients for Image/Video Processing

and Analysis. University of Buckingham.

Bicego, M., Grosso, E., & Tistarelli, M. (2005). Probabilistic face authentication using Hidden Markov

Blackburn, D., Bone, J., & Phillips, P. J. (2001). Face recognition vendor test 2000, . In: Defense Advanced

Research Projects Agency, Arlington, VA.