利用複合式多振幅限制離軸角頻譜演算法實現廣視角全像抬頭顯示器

Wide Field of View Holographic Head-up Display Realized by Hybrid Multi-Constraint Shifted Angular Spectrum Algorithm

程藝(Cheng, Yi)

指導教授 ： 黃乙白 ； 田仲豪

國立交通大學/電機學院/光電工程系所/碩士(2019年)

摘要

近年來平面顯示器技術已逐漸趨於成熟，人們開始追求更具有視覺衝擊的3D立體影像以及更具廣度的實用性顯示器，例如頭戴顯示器、抬頭顯示器、可撓顯示器等，並結合虛擬實境(VR)、擴增實境(AR)等技術提高真實性且達到能進行互動的影像效果。其中，常見的光場立體顯示是利用雙眼視差，即讓左右眼觀接收不同資訊疊加後在腦中建立的影像，但這會因為單眼對焦和雙眼對焦之間的差異使得人們在長時間觀看下經常碰到暈眩不適等狀況。然而，利用全像術所投影出的立體影像則不會有此問題產生。
由於全像術的基本原理是讓光確實打入雙眼來看到影像，因此不會有聚焦錯位的情況發生。在應用上，結合了深度資訊的AR全像車用抬頭顯示器可以讓影像以更自然的方式出提供駕駛即時路況資訊。不過全像投影伴隨著計算時間過長、視角太小等缺點，因此如何加速計算和擴大視角成為全像相關研究上的一大課題。
本論文探討基於離軸角頻譜演算法，藉由多振幅限制的方式加速生成擴大1.5至2倍視角且高品質之純相位式電腦生成全像片，並結合以子全像片為基礎的相位分割演算法於重建少數影像時，在相同FOV的情況下僅使用與計算單張影像相近的演算時間；在重建多張影像時則是延續相位分割法，在圖像有重疊的區域結合複數場疊加法形成的複合式相位多工演算法，不僅達到縮短約2至3倍計算時間的同時仍可維持良好的影像品質，且適用於多焦距成像，這對於往後在車用3D抬頭顯示應用上所要求的具深度、廣視角及即時更新等需求都能提供相當顯著的改善。

關鍵字

全像 ； 抬頭顯示器 ； 角頻譜演算法 ； 顯示器 ； 多振幅限制角頻譜演算法

並列摘要

In recent years, flatform display technology has gradually matured, and people start to pursue more visually stunning 3D stereoscopic images and practical displays with considerable breadths, such as head-mounted display (HMD), head-up display (HUD), flexible display…etc., which combined with virtual reality or augmented reality (AR) to improve authenticity and achieve interactive image effects. Among them, the light field technology that is commonly applied in the stereoscopic display is the use of binocular parallax, which may easily cause the vergence-accommodation conflict (V.A.C). However, the stereo image projected by the holography technique does not cause this problem.
The basic concept of holography is that the light of holography is actually converged and enter people’s eyes, which can completely eliminate the bad feeling from V.A.C. As for the application, automobile AR holographic HUD with depth information could allow the images to remind the drivers more naturally. Therefore, the short-coming of holographic display such as the time-consuming and narrow field of view (FOV) nowadays becomes a major topic in holographic research.
In this these, based on shifted angular spectrum algorithm, the multi-amplitude constraint method is proposed to accelerate the generation of high quality phase-only hologram with about 1.5 to 2 times FOV expansion, and combined with sub-hologram method (SHM) in few patterns reconstruction to obtain the same FOV in only the time similar to the calculation of a single image. For reconstructing multiple images, the use of hybrid phase multiplex (HPM) algorithm, which is proposed to fuse the SHM and complex field superimposing algorithm, not only can keep the high image quality with shortening 2 to 3 times computation time but also suitable in multi-focal imaging. These advantages provide a significant improvement in the depth, wide FOV, and real-time update requirements for automotive 3D HUD applications.

並列關鍵字

Holographic Head-up Display ； Head-up Display ； Shifted Angular Spectrum Algorithm

參考文獻

[1] A. Pauzie, “Head Up Display in Automotive: A New Reality for the Driver,” International Conference of Design, User Experience, and Usability, Springer, Cham, pp. 505-516, 2015.

[2] P. Iwaneczko, K. Jędrasiak, and A. Nawrat, “Augmented Reality in UAVs Applications,” Innovative Simulation Systems, pp. 77-86, 2016.

[3] Y. Hwang, B. Park, and K. Kim, “Effects of Augmented-Reality Head-up Display System Use on Risk Perception and Psychological Changes of Drivers,” ETRI Journal, vol. 38, no. 4, pp. 757-766, 2016.

[4] N. Broy, S. Höckh, and A. Frederiksen, “Exploring Design Parameters for a 3D Head-Up Display,” Proceedings of The International Symposium on Pervasive Displays, ACM, p. 38, 2014.

[5] J. Christmas, D. Masiyano, and N. Collings, “Holographic Automotive Head Up Displays,” Electronic Displays Conference, 2015.