Full metadata record
DC FieldValueLanguage
dc.contributor.author朱允維zh_TW
dc.contributor.author吳炳飛zh_TW
dc.contributor.authorChu, Yun-Weien_US
dc.contributor.authorWu, Bing-Feien_US
dc.date.accessioned2018-01-24T07:42:44Z-
dc.date.available2018-01-24T07:42:44Z-
dc.date.issued2017en_US
dc.identifier.urihttp://etd.lib.nctu.edu.tw/cdrfb3/record/nctu/#GT070460065en_US
dc.identifier.urihttp://hdl.handle.net/11536/142849-
dc.description.abstract駕駛之生理狀態攸關了行車與用路人安全,而心跳為判斷人健康狀態之重要指標,若能長期監控駕駛心跳不僅保障駕駛人健康也確保交通安全。傳統心跳儀器多利用接觸式裝置量測,且需將皮膚緊貼感測器,應用於駕駛心跳監控恐會造成駕駛不適,或因為皮膚無法持續接觸感測器而使量測失準,近年來開始有遠距影像式心跳量測之技術,利用攝影機連續拍攝影像,透過偵測受測者臉部之光源變化而得到其心跳頻率,但此技術之準確率與環境光、駕駛臉部之受光狀況極為相關,多數研究將實驗場景設定為可控制之室內,戶外多變光源場景於駕駛臉部造成之光影變化也較少受到討論。因此本論文目標為使實驗環境為戶外之真實道路駕駛,並將多種天候、不同時間所造成的多變環境光源納入考量,開發一套可抵抗駕駛臉部多變光源差異之即時心跳量測系統。 本論文首先透過長期蒐集駕駛於平日之影像與心跳資料庫,根據不同的心跳區間訓練多個神經網路模型,並透過改良人臉偵測,以提高駕駛於開車時臉部多角度之偵測準確率,由臉部感興趣區域取出時域心跳訊號並濾波後,於頻譜中取出頻域特徵並考慮頻域之穩定度,當頻域混亂時將預先訓練之個人化神經網路模型應用至心跳之預測,以克服駕駛臉部光影變化時於頻域所產生之雜訊。演算法除考慮不同天候與光影變化下可見光環境之駕駛心跳偵測外,也將應用拓展至低光源之夜晚,於全天候且不同時間下量測駕駛之心跳。zh_TW
dc.description.abstractDriver’s physiological status can be of enormous value to public traffic safety and cannot be ignorable nowadays. Additionally, heartbeat is one of the most important indicators of human’s health status. To detect driver’s heart rate, traditional contact-type devices might bring about driver’s distraction and discomfort. Remote PhotoPlethysmoGraphy (rPPG) technique is suitable for vehicle application to monitor driver s’ heartbeat by a web camera without interfering drivers. Most of the rPPG research intended to enhance the robustness of facial motion or luminance change in indoor or controllable scenario, but outdoor or realistic scenarios have generated relatively little discussion. Consequently, the purpose of this paper is to enhance rPPG technique that suitable for the outdoor driving vehicles scenario, for monitoring driver’s heart rate in different weather condition even in daytime or night. Collecting driver’s long-term continuous images and heartbeat information, we customize personal ANN multi-model for each driver. Through determining entropy of the spectrum and adaptively utilizing personal ANN model, proposed algorithm eliminates the effect of noise caused by the difference of facial luminance in outdoor vehicle scenarios. The proposed system also consider the full dark scenario, monitoring driver’s heart rate at different times. This study provides a promising solution for rPPG technique on driver’s heartbeat monitoring and expects to improve driving safety.en_US
dc.language.isozh_TWen_US
dc.subject先進駕駛輔助系統zh_TW
dc.subject類神經網路zh_TW
dc.subject心跳間控zh_TW
dc.subject影像式心跳偵測zh_TW
dc.subject駕駛安全zh_TW
dc.subjectAdvanced Driver Assistance Systemsen_US
dc.subjectArtificial Neural Networken_US
dc.subjectHeart Rate Monitoringen_US
dc.subjectRemote Photoplethysmographyen_US
dc.subjectVehicle Safetyen_US
dc.title基於類神經網路之抗光影影像式心跳偵測應用於多天候駕駛場景zh_TW
dc.titleANN and Luminance Robust Image Based Photoplethysmography for Drivers’ Heart Rate Monitoring for All-weather Scenariosen_US
dc.typeThesisen_US
dc.contributor.department電控工程研究所zh_TW
Appears in Collections:Thesis