利用時間和空間多工降低AR-HUD之光斑

Abstract

基於LCoS(liquid crystal on silicon)與全像多色投影之雙視距車用抬頭顯示器(Dual-focal-plane Multi-color Head Up Display with Holographic Imaging)[1]研究中，利用雷射當作光源而嚴重影響到成像品質的光斑(speckle)問題，導致駕駛無法迅速掌握路況或是儀表板資訊，而影響到整體駕駛體驗，甚至是行車安全，此問題與研究的初衷和原本想帶給駕駛的行車體驗背道而馳。本論文使用光斑對比度(speckle contrast, SC)的參數作為全像片光斑問題的依據，藉由量測光斑對比度之數值來判斷全像片的優劣。 藉由上個實驗[1]所使用的空間多工混色可以有效降低光斑的方法，搭配時序法的方式，採用“時間和空間多工法”來降低光斑問題。一般時間多工法是利用單波長雷射光源搭配一片投影面板來投影並且快速切換全像片，但這樣除了需要三台投影面板外，面板的響應頻率會直接影響到結果，而空間多工法因為在同一全像片上分配多個波長的相位，導致單個相位所使用到的pixel數量下降，進而導致解析度下降，所以在本實驗中經由將三個不同波長的雷射儀投射至同一面板上，這樣可以將架構體積縮小外，配合空間多工法來達到全彩投影，而因為空間多工法的關係，我們可以將單張全像片投影出兩個波長相位的全像圖，以此降低面板響應頻率的要求，就可以達到時間多工法的效果，所以本實驗架構是由時間多工法以及空間多工法相輔相成、取兩個方法各自的優點來進行實驗，以此更有效率的降低光斑問題。 文中我們將介紹光斑現象和原理，接著介紹實驗中的全像投影系統和環境設定。在我們的系統中，全像圖以MGSA(modified Gerchberg–Saxton algorithm)的方式來製作，將雷射儀與LCoS連結至電腦來控制雷射光和全像圖的切換頻率達到一致，圖像在擴散片(diffuser)上接收圖像，在光路中間加上截波器(chopper)配合電腦播放全像圖來達到播放指定波長全像圖同時對應波長的雷射光達到時序法，並利用固定的焦距和光圈大小的相機拍攝投影畫面，比較圖像的光斑對比度，而在計算光斑對比度的過程中，我們只取圖片中雷射光最密集的某一個部分作分析，使照片中非圖像的部分不會影響到最終對比度的結果。經由測量後，我們最終可以得知，藍色單張全像片光斑對比度為28.86%、綠色為22.22%、紅色為14.68%，而本實驗所使用的“時間和空間多工法”光斑對比度為4.51%，達到有效降低光斑的效果。

In the research on Dual-focal-plane Multi-color Head Up Display with Holographic Imaging based on LCoS (liquid crystal on silicon) and holographic projection, the issue of speckle caused by using lasers as light sources seriously affects image quality. If the speckle is too large, it can result in insufficient image resolution or blurry image edges, making it difficult for drivers to quickly grasp road conditions or dashboard information. This can negatively impact the overall driving experience and even driving safety, which contradicts the original intent of the research.

By utilizing the spatial multiplexing color mixing method employed in the previous experiment [1], combined with a time multiplexing method, called "time and spatail multiplexing" to address the speckle issue effectively.Time multiplexing methods using a single-wavelength laser source in conjunction with a projection panel, and rapidly switching between holograms. However, this method requires three projection panels and the panel's response frequency significantly affects the results. On the other hand, the spatial multiplexing method distributes multiple wavelength phases on the same hologram, leading to a reduction in the number of pixels used for each individual phase and thus a decrease in resolution.

In this experiment, three lasers of different wavelengths are projected onto the same panel. This not only reduces the overall system size but also achieves full-color projection through spatial multiplexing. Additionally, due to the nature of spatial multiplexing, it becomes possible to project two-wavelength phase holograms from a single hologram plate, reducing the panel's response frequency requirements and effectively achieving the temporal multiplexing effect.

Therefore, this experimental setup combines both temporal and spatial multiplexing methods, harnessing the advantages of each to efficiently address the speckle issue.The paper will introduce the phenomenon and principles of speckle, followed by an introduction to the holographic projection system used in the experiment and the environmental settings. Speckle contrast (SC) is introduced as a crucial parameter for comparing speckle size.

We will then present conventional methods for reducing speckle and the "half-space, half-time" approach used in this study. In our system, holograms are generated using the modified Gerchberg-Saxton algorithm (MGSA), and lasers and the LCoS are connected to a computer to control the switching frequency of laser light and holograms to match. Images are captured on a diffuser screen with the addition of a chopper in the optical path, synchronized with the computer's playback of specified wavelength holograms and corresponding laser light to achieve temporal multiplexing. A fixed-focus camera with a specific aperture size is used to capture the projected images, and speckle contrast is calculated by analyzing a specific region with the densest laser light in the photos, ensuring that non-image areas do not affect the final contrast results.