基於巨觀邊緣感知與對比圖分析的高動態範圍成像

dc.contributor高文忠zh_TW
dc.contributorKao, Wen-Chungen_US
dc.contributor.author鄭翔元zh_TW
dc.contributor.authorCheng, Hsiang-Yuanen_US
dc.date.accessioned2025-12-09T08:31:20Z
dc.date.available2025-08-14
dc.date.issued2025
dc.description.abstract本研究主要探討高動態範圍影像處理中,在極端光照條件下的影像主體重建與色調映射問題。當畫面遭遇強烈背光、低光源或大面積背景干擾時,傳統的影像處理方法經常無法有效地凸顯主體,且在壓縮動態範圍過程中容易造成細節喪失與色彩失真。本研究針對上述問題,提出一套考慮巨集邊緣資訊的影像處理方法,結合全域與區域對比度評估,透過調整適合的色調映射曲線,使暗部細節清晰且避免亮部過曝。此外,研究中亦針對色彩還原問題,於色彩轉換過程中引入色域映射模型的補償機制,有效避免傳統方法常見的色相偏移與飽和失真現象。本論文透過實驗驗證所提出方法的有效性,並經由業界常見的客觀指標評估其在亮度、色彩準確性與視覺對比度上的改善效果,期望能提供未來影像訊號處理系統設計的重要參考。zh_TW
dc.description.abstractThis study addresses subject reconstruction and tone mapping in HDR images under extreme lighting—strong backlight, low illumination, or large-area distractions. We propose a framework that leverages macro-edge information and combines global and local contrast evaluation to adapt tone-mapping curves, preserving shadow detail without over-exposing highlights. To improve color fidelity, an OKLab-based compensation is applied during color conversion, reducing hue shifts and saturation errors. Experimental results and VCX metrics confirm notable gains in luminance reconstruction, color accuracy, and visual contrast, offering a practical reference for future ISP system design.en_US
dc.description.sponsorshipAI跨域應用研究所zh_TW
dc.identifier612K0020C-48366
dc.identifier.urihttps://etds.lib.ntnu.edu.tw/thesis/detail/71e44b4e5a49c1f02f5c0dde7a89bbd8/
dc.identifier.urihttp://rportal.lib.ntnu.edu.tw/handle/20.500.12235/126198
dc.language中文
dc.subject高動態範圍zh_TW
dc.subject色調重現zh_TW
dc.subject色域映射zh_TW
dc.subject對比增強zh_TW
dc.subjectHDRen_US
dc.subjectTone Reproductionen_US
dc.subjectTone Mappingen_US
dc.subjectContrast Enhancementen_US
dc.title基於巨觀邊緣感知與對比圖分析的高動態範圍成像zh_TW
dc.titleHigh Dynamic Range Imaging via Macro-Edge Awareness and Contrast Map Analysisen_US
dc.type學術論文

Files

Original bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
202500048366-110691.pdf
Size:
8.48 MB
Format:
Adobe Portable Document Format
Description:
學術論文

Collections