Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/48949
標題: Design and Implementation of Intelligent Baby Facial Expression Recognition and Foreign Body Detection
智慧型嬰幼兒監護與安撫晶片系統-子計畫二:智慧型嬰幼兒臉部表情辨識與臉部異物偵測設計與實作
作者: 范志鵬
關鍵字: Baby Watch and Comfort
嬰幼兒監護與安撫
電子電機工程類, 資訊科學軟體
Intelligent
Face Feature
Facial Foreign Body Detection
FacialExpression Recognition
High Speed
System-on-Chip Realization
智慧型
臉部特徵
臉部異物偵測
臉部表情辨識
高速
嵌入式系統晶片實現
應用研究
摘要: In this project, the main purposes are researches and developments of intelligent high-speedbaby facial expression recognition and baby facial foreign body detection technologies by automaticschemes for indoor baby watch and comfort. The developed algorithms can automatically find thebaby facial position, and then extract the baby facial features for facial expression recognition nomatter how the background is complex or pure, the light is enough, and the sizes of baby faces aredifferent. Thus, the main contents of the research include the baby facial position detection, thebaby facial feature extraction, the baby facial expression recognition, and the baby facial foreignbody detection. In the part of the baby facial position detection, we will apply the evolution anddeduction algorithm to solve this problem. By detecting the skin color and using the oval-shapedfacial model, the correct facial position will be calculated and detected. In the part of the babyfacial feature extraction, based on the results of the facial position detection, we will extract thefacial features by using the related positions of the features and the facial geometry model, wherethe facial features include mouth, eyes, eyebrows, and nose. Then we further extract these featurepoints automatically for the recognition of the facial expressions. In the part of the baby facialexpression recognition, we will apply the intelligent fuzzy algorithm and the feature-point vectordistances to analyze the facial expression, and then the baby facial foreign body detection also willbe done, simultaneously. The three-year researches of this project are divided into two topics asfollows.(1) Intelligent baby facial foreign body detection and its embedded SOC realization :This research topic will be focused on the high-speed and intelligent baby facial foreign bodydetection. The applications are the monitor and watch of babies.(2) Intelligent baby facial expression recognition and its embedded SOC realization :This research topic will be focused on the high-speed and intelligent baby facial expressionrecognition. The applications are the watch, the comfort, the entertainment, and the education ofbabies.
在所有嬰幼兒外在反應中最重要的就是其臉部表情的分析,因為情緒及生理狀態反應最明顯的外觀就是在臉部表情。所以研究嬰幼兒臉部表情的分析及嬰幼兒的臉部偵測,使嵌入式晶片系統能智慧型且高速的了解嬰幼兒的臉部狀態,以便對嬰幼兒的狀況提出最快及最佳的反應及警示,是非常重要的。而發展智慧型嬰幼兒監護與安撫的臉部表情辨識與臉部異物偵測技術,可有效地節省監控與照顧的人力,並且提高監護與安撫的素質,快速反應意外或危急的狀態,以減少遺憾的發生。對嬰幼兒而言,亦可提供安撫,娛樂與教育的功能,使嬰幼兒與嵌入式晶片系統進行影音的互動學習,以減輕照顧者的負擔。本子計畫預計研究出一套高速,智慧型,且自動化的嬰幼兒臉部表情辨識與臉部異物偵測系統,在不限定背景顏色是否單純,不限定光線是否足夠,且不限定影像中嬰幼兒臉部大小的情況下,從影像進來就可自動化的正確找出嬰幼兒臉部位置,然後擷取嬰幼兒臉部特徵進行表情辨識。故研究內容主要包括嬰幼兒臉部位置偵測、特徵擷取、表情辨識以及異物偵測四個部分。在嬰幼兒臉部偵測的部分,將計畫採用進化演繹法,利用皮膚顏色及橢圓形狀來正確的偵測出人臉的位置,並準確的圈出臉部。在特徵的擷取的部分,根據人臉偵測的結果,利用相對的位置與人臉幾何模型來擷取特徵,包含嘴巴、眉毛以及眼睛,然後再進一步自動的取出這些部分的特徵點,作為判斷表情的依據。在表情辨識的部分,將採用模糊智慧型演算法,將前步驟找出的特徵點分析處理後,判斷是屬於何種表情,並且同時做臉部異物的偵測。 本三年度的子計畫研究方向可分為(1) 智慧型室內照顧的嬰幼兒臉部異物偵測技術研究與嵌入式系統晶片實現 :在於高速且智慧型的室內照顧臉部異物偵測,以完成嬰幼兒的監控與照顧問題。在此技術的研究設計上,高速反應,智慧型,自動警告與通知的是設計的三大目標。(2) 智慧型室內照顧的嬰幼兒臉部表情辨識技術研究與嵌入式系統晶片實現 :在於高速且智慧型的室內照顧臉部表情辨識,以達成嬰幼兒的監控/照顧/娛樂/教育等的功能。在此技術的研究設計上,高速反應,智慧型,自動化互動是設計的三大目標。
URI: http://hdl.handle.net/11455/48949
其他識別: NSC97-2221-E005-088-MY2
Appears in Collections:電機工程學系所

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.