人工智慧影像優化之屏下顯示超穎光學系統( I );Ai Image Optimization of under Display Meta-Optic-System( I )

NCU Institutional Repository > 研究中心 > 光電科學研究中心 > 研究計畫 > Item 987654321/84579

請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/84579

題名:	人工智慧影像優化之屏下顯示超穎光學系統( I );Ai Image Optimization of under Display Meta-Optic-System( I )
作者:	王智明;李建階;孫慶成;陳昇暉
貢獻者:	光電科學研究中心
關鍵詞:	超穎材料;超穎介面;電漿子學;超穎透鏡;屏下鏡頭;介電超穎表面;幾何相位;機器視覺;卷積神經網絡;metamaterial;metasurface;plasmonics;metalens;under-screen camera;all-dielectric metasurface;geometric phase;machine vision;Convolutional Neural Networks
日期:	2020-12-08
上傳時間:	2020-12-09 09:54:26 (UTC+8)
出版者:	科技部
摘要:	沒有按鍵、邊框或瀏海的全屏幕智慧手機已逐漸成為手機市場的主流。對於全屏幕智慧手機而言，現有放在前屏幕的裝置都得隱藏在屏幕後面，包括前置鏡頭，指紋感測器等。例如，蘋果公司已經獲得了基於聲波成像系統的屏幕下指紋感測器的專利，用戶可直接觸摸屏幕上的任意位置來解鎖iPhone。中國智慧手機製造商小米和OPPP也提出了隱藏在屏幕後面的前置鏡頭。儘管首款配備屏幕下攝像頭的智能手機無法提供最佳的圖像品質。但對於手機行業來說，這仍然是必須要發展的項目。本計畫旨在為基於超穎光學技術的屏下成像系統提供完整的解決方案。由於屏下光學成像系統的總長受限於筆記本電腦和以及智慧手機的有限厚度，因此，平面型超薄透鏡相對適合此應用。在製程方面，我們將以基於濺鍍製程的GaN 製作全介電超穎表面的消色差PB-phase超穎透鏡並將其封裝在CMOS感測器上，應用於屏幕前數十公分的物體的屏幕下取像。此外由於屏下影像會受到於散射、色偏，遮蔽等影響，為了提高屏下影像的影像品質，我們也將建立一個深度3D卷積神經網絡（3D CNN）模型和訓練數據庫。根據上述目的，這個為期三年的項目分為三個子項目：屏幕下成像系統設計：從幾何光學、波動到電磁波分析。低成本的氮化物PB-phase超穎透鏡製程技術； Al影像優化：數據庫和訓練模型。最後也將結合以上三子題成果，提供一個屏下顯示技術的完整解決方案。 ;Full-screen and all-screen smartphones, with no buttons or borders to interrupt the pure design, have gradually become the mainstream of the cell phone market. For all-screen smartphones, all of the devices should be hidden behind the screen, including front cameras, fingerprint sensors and so on. For example, Apple has received a patent for an under-screen fingerprint sensor based on an acoustic imaging system that would allow users to unlock an iPhone by touching anywhere on the screen. Chinese smartphone makers, Xiaomi and OPPP, also proposed the front-facing camera hidden behind the screen. Although the first range for smartphones with under-screen cameras won’t deliver the best image quality around. It is still a must-do for cell phone industries. This project aims to deliver a total solution for the under-screen camera based on meta-optics technology. The total track of the optical imaging system is limited by the finite thickness of a laptop and a smartphone. Therefore, a flat lens is essential for this application. Here, we will deliver chip-bonding achromatic PB-phase metalens on CMOS sensors for under-screen imaging. The PB-phase metalens, which is a GaN all-dielectric metasurface, is utilized to capture the image of an object with tens centimeters in front of a screen. For improving the image quality of the captured under-screen images suffering from scattering, color rendering, shadow, and so on, a deep 3D Convolutional Neural Network model and training databased will be established as well. According to the above-mentioned purpose, this three-year project is divided into three sub-project: Design of under-screen imaging system: from macro-, micro- to meta-optics; Nanofabrication of low-cost nitride-based metalens on the glass; Al image optimization: database and training model.
關聯:	財團法人國家實驗研究院科技政策研究與資訊中心
顯示於類別:	[光電科學研究中心] 研究計畫

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	245	檢視/開啟

在NCUIR中所有的資料項目都受到原著作權保護.

社群 sharing

資料載入中.....