請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/24176
標題: | 基於主動外觀模型之即時表情辨識系統應用於人機互動 Realtime Recognition of Facial Expressions Based on Active Appearance Model for Human-Robot Interactive Applications |
作者: | Chun-Yen Huang 黃俊諺 |
指導教授: | 羅仁權 |
關鍵字: | 臉部特徵的匹配與追蹤,主動外觀追蹤模型(AAM),光流法,臉部表情辨識,人機互動, Alignment and Tracking of Facial Features,Active Appearance Model (AAM),Optical Flow,Facial Expression Recognition,Human-Computer Interactions, |
出版年 : | 2011 |
學位: | 碩士 |
摘要: | 站在二十一世紀的開端,承續二十世紀人類文明輝煌的進展,目前人類生活的重心已經不再是生存和延續生命這些基本的層面,而是在於追求更精緻的生活品質和更輕鬆的生活方式。醫學和藥品科技的進步也使得人類平均年齡大幅提高,徹底改變世界人口結構,隨著老年人口急遽增加,社會福利、醫療照護和居家看護陪伴等等所需要的投入資源也相對提昇。基於社會上這樣的需求,機器人的應用因此開始蓬勃發展,從工業上的用途開始逐漸進入人們的生活中。也因為如此,智慧型機器人的發展已經是刻不容緩,成為世界各國爭相投入發展的領域,表情辨識也因此成為人機互動的領域中一個十分重要的課題。
我們研究的主題在於智慧型機器人利用攝影機取得影像與周遭環境產生互動,包含使用主動外觀模型(Active Appearance Models, AAM)去追蹤使用者的臉部特徵,和利用使用者臉部特徵的資訊去判斷使用者的臉部表情,並將這樣的系統應用在機器人或電腦上來做進一步的分析。根據這樣的資訊,機器人或電腦便可以輕鬆的偵測出使用者的臉部表情和情緒的變化,並據此做出適當的回應。好的AAM校準結果需要有適當的AAM模型起始位置,而當我們在AAM上套用影像金字塔,也需要大量的運算時間才能得到精準的結果。在本篇論文中,將會詳細的介紹AAM的原理以及一個新的演算法來解決上述出現的問題。在我們使用的演算法中,應用了元件式主動外觀模型(Component-based AAM, CB-AAM ),並使用左眼、右眼、嘴巴的子模型來分別進行匹配。藉由這樣的方式,可以使得AAM匹配臉上特徵的速度更有效率,也使得整個演算法有能力去進行臉部特徵的即時追蹤,處理現實世界中攝影機收到的即時影響。為了要得到更穩定的匹配結果,我們使用了包含多階影像金字塔的光流法來決定AAM模型放置的起始位置,這除了使得AAM的匹配結果更穩定,也讓整個系統花費更少的疊代次數就完成足夠精準的匹配結果。 使用我們所發展的演算法可以相對快速且穩定的得到使用者的臉部表情資訊,並取得使用者臉部特徵的詳細資訊以應用在其他更進一步的應用上。 At the beginning of the 21st century, the advancement of medical and pharmaceutical technology also cause to the average age of human increase significantly, and thus increasing resources needed in social welfare, medical care, and home care. According to the need, the application of robots starts growing vigorously and joining people’s daily lives. As a result, lots of researchers around the world devote to the developing of intelligent robots. Recognition of facial expressions becomes one of the significant issues between human and robot interactions. In daily lives, we need more methods to control our robots, even without using hands. And we need robots to watch for user’s need even before user give out commands. The objective of my research is to implement a realtime facial feature tracking system and recognize user’s facial expression with the system. The content of this thesis is to study the alignment and tracking of facial features with optical flow and component-based active appearance model, and then analyze fitted points to recognize facial expressions. Using this method with accurate analysis and tracking of facial features in robot or computer. Consequently robot or computer can easily recognize user’s facial expressions and emotional variation, and then response properly. We apply some realtime techniques and Active Appearance Model (AAM) on the cameras. A high-quality AAM alignment results depend on appropriate selections of initial positions. Nevertheless it takes a lot of time when we apply image pyramid to get precise results. In this paper, we introduce a new method to apply AAM fitting and further solving above problems. In our fitting plan, we apply component-based AAM fitting with AAM sub-model fittings separately on mouth and eyes. Therefore, we could make more efficient facial features alignment and then it becomes able to implement tracking to real-world video and realtime alignment. To get more stable fitting results, we use multi-level optical flow to determine initial positions of facial feature models, and this makes our system able to complete AAM fitting within much less iterations. Our tracking method has good abilities of tracking moving front -face and changing facial expression, and our recognition system is able to recognize user’s facial expression in normal environment with realtime speed. It is relative easier to analyze user’s emotional information and get accurate positions of facial features for further application in real world environments by the algorithm we developed. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/24176 |
全文授權: | 未授權 |
顯示於系所單位: | 電機工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-100-1.pdf 目前未授權公開取用 | 2.92 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。