基於AAM的自動化人臉表情辨識系統

Automatic Facial Expression Recognition System Based on Active Appearance Model

林晉頤(Chin-I Lin)

指導教授 ： 吳明川

國立臺北科技大學/機電學院/自動化科技研究所/碩士(2008年)

https://doi.org/10.6841/NTUT.2008.00629

摘要

本研究係利用個人電腦，研製一套可以在數位影像中自動搜尋所有人臉及定位臉部特徵的表情辨識系統。在表情辨識的課題中，臉部特徵定位是一項關鍵步驟；但是現實環境存在許多不利因素，可能使臉部特徵定位失敗導致無法有效的判斷表情，例如膚色差異、姿態變化、異物遮蔽臉部等等。我們嘗試引入主動外觀模型（Active Appearance Model）以解決上述特徵定位問題。
我們依據表情辨識的需求，選定64個臉部特徵點及163張人臉樣本來訓練AAM。為了強化特徵定位性能，本研究更結合Adaboost Cascade Face Detection人臉偵測演算法並設計一套定位調校流程，以應用在多人臉情況。最後我們建立607筆表情資料，並將定位完成的特徵資訊經由簡單的K鄰近分類演算法做分類。在網路下載之179張人臉表情分類實驗中達到不錯的辨識率。

關鍵字

表情辨識 ； 特徵定位 ； 主動外觀模型 ； 主動影像調校

並列摘要

This research is to develop a facial expression recognition system that is able to automatically detect all faces and position facial features in digital images in the personal computer environment. Feature position is a key of the steps in the expression recognition task, but many unfavorable factors in reality, such as skin color difference, pose variance, obscured face, etc, may cause position failure and invalid expression result. In this study, we apply Active Appearance Model to solve this problem.
We consider 64 landmarks and 163 face images to train the model for expression recognition. In order to enhance the performance of feature position, meanwhile we apply Adaboost Cascade Face Detection and design a set of adjusting processes for multiple-face cases. Finally, we make a realization of expression recognition by 607 expression samples easily using K-Nearest Neighbor algorism to classify the facial features which have been located, and have a good recognition rate in the experiment of 179 face pictures gathered from the Internet.

並列關鍵字

Facial Expression Recognition ； Feature Position ； Active Appearance Model ； Active-Image Adjustment Processes

參考文獻

[14] 林建成，人臉表情自動辨識系統之研製，碩士論文，國立台北科技大學，2006。

[2] 吳明衛，自動化臉部表情分析系統，碩士論文，國立成功大學資訊工程系，2004。

[3] S. Birchfield, "Elliptical head tracking using intensity gradients and color histograms," IEEE International Conference on Computer Vision and Pattern Recognitions, Santa Barbara, California, 1998, pp.232-237.

[4] S. Birchfield, "An elliptical head tracker," Proceedings of the 31st Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, California, 1997, pp. 1710-1714.

[9] Rein-Lien Hsu, Mohamed Abdel-Mottalab, and Anil K. Jain, "Face Detection in Color Images," IEEE Trans. Pattern Analysis and Machine Intelligence, vol.24, no.5, 2002.