運用臉部器官之多特徵性別分類

Component-based Multi-Features Facial Gender Classification

劉俊豪(Chun-Hao Liu)

指導教授 ： 陳文雄

國立暨南國際大學/科技學院/電機工程學系/碩士(2012年)

摘要

近年來，生物資訊蓬勃發展，尤其是臉部器官特徵更是熱門的研究領域。早期的全自動人臉性別辨識系統受限於不成熟的人臉偵測技術，近年來人臉偵測技術逐漸地成熟，因此人臉性別辨識也開始受到重視，並可應用於廣告和監控領域中。
本論文提出利用臉部器官特徵的兩階段性別分類，在第一階段利用男性獨有的特徵，例如：禿頭和鬍鬚，使得能在第一階段有效地篩選出男性，接下來在第二階段，利用主動形狀模型（Active shape models, ASM）擷取人臉五官輪廓，快速執行臉部特徵點定位，將人臉五官及人臉輪廓幾何特徵透過支持向量機（Support vector machine, SVM）進行辨識男女性別。整個分類系統主要朝向三個方向來進行實作：影像前處理、特徵萃取與分類辨識。
本論文採用FEI人臉資料庫、FERET人臉資料庫、真實環境下拍攝照片，作為測試資料庫，分別從中取出正面臉部圖片160張（男女各80張）、1900張（男女各950張）120張（男女各60張）人臉影像進行實驗測試，最好的性別辨識率可達到88.125%。

關鍵字

生物辨識 ； 人臉偵測 ； 人臉辨識 ； 性別分類 ； 毛髮特徵 ； 臉部特徵

並列摘要

Along with the development of biological information was growing rapidly in recent years, the technology of component-based feature has also become a popular research area. The early fully-automatic system of face gender recognition was quite limited by the immature technology. However, the face detection techniques have been gradually getting precise and reliable in recent years, so the study of face gender recognition was respected again. There are many applications which require the technology, such as advertising or monitoring.
In this thesis, a two-stage component-based face gender classification scheme is presented. Since some of the specific features, such as baldness, mustache and goatee, appear obviously only on a male human face, this scheme will filter those male images with the unique features out at the first stage. In the second stage, we used Active Shape Model (ASM) to fast locate face feature points, the feature regions would be accurately extracted and recognized by Support Vector Machine (SVM) that accomplished the gender recognition. This classification system includes three modules: image pre-processing module, feature extraction module, and classification module.
In this thesis, the analysis database is adopted from FEI face database, FERET face database and real photos. We separately select the sample amount of 160(half males and half females), 1900(half males and half females) and 120 (half males and half females) that are the face photographs to analyze. The best performance of gender recognition rate is 88.125%.

並列關鍵字

Biometrics ； Face Detection ； Face recognition ； Gender classification ； Hair Feature ； Face Feature

參考文獻

[1].H. A. Rowley, S. Baluja, and T. Kanade, “Neural Network-Based Face Detection,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 20, no. 1, pp. 23-38, Jan. 1998.

[2].H. A. Rowley, S. Baluja, and T. Kanade, “Rotation Invariant Neural Network-Based Face Detection,” in Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 38-44, 1998.

[3].K.K. Sung and T. Poggio, “Example-Based Learning for View-Based Human Face Detection,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 20, no. 1, pp. 39-51, Jan. 1998

[4].R. F´eraud, O.J. Bernier, J.E. Viallet, and M. Collobert, “A Fast and Accurate Face Detection Based on Neural Network,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 23, no. 1, pp. 42-53, Jan. 2001.

[5].R.L. Hsu, M. Abdel-Mottaleb, and AK. Jain, "Face Detection in Color Images," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 24, no. 5, pp. 696-706, 2002.