結合人眼萃取模型之主動式車用 遮陽板之研究;Study on active vehicle sun visors incorporated with human-eye extraction model

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/95019

Title:	結合人眼萃取模型之主動式車用遮陽板之研究;Study on active vehicle sun visors incorporated with human-eye extraction model
Authors:	楊宗翰;Yang, Zong-Han
Contributors:	光電科學與工程學系
Keywords:	眩光防護;機器學習;影像辨識;液晶顯示器;Glare protection;Machine learning;Image recognition;Liquid crystal display
Date:	2024-08-13
Issue Date:	2024-10-09 15:43:50 (UTC+8)
Publisher:	國立中央大學
Abstract:	本論文設計了一套可應用於車載之防眩光系統，且考慮車載之實時應用我們提出了低演算成本之方法實現此系統，而本系統為模擬駕駛駕車時之情形，使用液晶面板、投射燈、攝影機及人眼模型等進行實驗。而我們將投射燈作為眩光來源並調製液晶局部之穿透率，並以攝影機拍攝人眼模型進行分析，使液晶面板能保護人眼模型不被眩光，而本系統可針對投射燈進行動態防護，且當眩光源消失時，系統也能偵測人眼模型中瞳孔與虹膜之區域眩光點消失，並停止防護之動作。在本論文中考慮了此系統應用於真實人眼之情形，因此建立了人眼萃取模型，且此模型可精準的萃取人眼中虹膜與瞳孔之區域，之後分析若可基於人眼萃取模型結合更具穩健性之眩光判別演算法，可實現應用於車載之實時防眩光系統。 ;This paper designs an anti-glare system for automotive applications. Considering the real-time use in vehicles, we propose a low computational cost method to implement this system. The system simulates driving conditions, using an LCD panel, a projector lamp, a camera, and an eye model for experiments. We use the projector lamp as the glare source and modulate the local transmittance of the LCD. The camera captures the eye model for analysis, allowing the LCD panel to protect the eye model from glare. The system dynamically protects against the projector lamp, and when the glare source disappears, it detects the disappearance of glare spots in the pupil and iris regions of the eye model and stops the protection action. In this paper, considering the application of this system to real human eyes, an eye extraction model was established. This model can accurately extract the iris and pupil regions of the eyes. Further analysis shows that combining this eye extraction model with a more robust glare detection algorithm can realize a real-time anti-glare system for automotive use.
Appears in Collections:	[Graduate Institute of Optics and Photonics] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	25	View/Open

社群 sharing

Loading...