藉由降低干擾提升立體影像品質於立體顯示器應用研究

Abstract

隨著顯示器發明以來，顯示器在我們是常生活中扮演不可或缺的角色。近幾年來立體顯示器技術成為顯示技術演進中的重點發展項目之一，除了可以提供觀賞者更為真實的體驗外，立體顯示技術更加影響了娛樂事業的發展。然而由於立體技術的蓬勃發展，對於其缺點「重影」需更加倍被重視，因為重影的現象大大的降低了立體影像的品質，並造成觀賞者不舒適感。至今雖然重影被視為最主要影響立體影像品質的因素，但根據文獻以及我們的實驗，我們認為重影並不能完全呈現雙眼視差式立體影像的品質。因此於此論文中，首先我們致力於藉由降低重影現象來改善立體影像，接著並找尋更適當的指標來評斷立體影像品質。 在此論文中我們提出三種方法來改善影像品質，其中包含了硬體與軟體方法。在硬體方面我們提出高阻值液晶透鏡應用於立體顯示技術來改善液晶透鏡的立體顯示效果，高阻值液晶透鏡有以下優點：平面與立體影像切換、低驅動電壓、簡易的結構與電路設計以及產生可接受的重影值。 而在軟體方面，首先我們提出數位式重影消除方法，於偏振式立體顯示技術中，在正負三十度的水平視角內重影可被降低至5%以下。而於裸眼顯示器應用中，重影值幾乎都會高於20%，而數位式重影消除方法於某些高度重影情況下會出現無法有效消除的情況。因此我們對於高度重影的顯示器提出較有效率的解決方法：分區式數位式重影消除方法。 根據文獻以及我們的人眼感知實驗中，我們發現立體影像深度以及動態影像模糊程度也會影像到立體影像品質。因此我們提出立體影像指標(∆X3D)來評斷立體影像，此指標包含了重影、影像深度以及動態影像模糊程度的影響。我們實驗結果顯示立體影像指標與人眼感知有高度正向關，相關係數達到0.8。最後本論文將立體影像指標拿來評斷我們提出的改善影像品質的方法，結果顯示不管是數位式重影消除方法還是分區式數位式重影消除方法皆可以有效提升立體影像品質。

Ever since displays were invented, they have become an essential electronic instrument in our daily lives. A natural evolution of these systems, i.e., 3D display technology, is one of the most promising display developments in recent years; it not only provides immersive and realistic experiences to the audience but also has significant influence on the entertainment industry. Nevertheless, the crosstalk issue in 3D technologies still needs to be further addressed. The images with crosstalk will substantially lower the image quality, and cause an uncomfortable feeling. Although crosstalk was regarded as the primary factor affecting 3D dynamic image quality, we believe that crosstalk can’t fully represent the human perception on the 3D image. Therefore, reducing crosstalk and evaluating 3D image becomes the main subject in this dissertation. In this dissertation, three methods had been proposed to improve 3D image quality. For hardware solution, we proposed high-resistance LC (Hi-R LC) lens to enhance the performance of LC lens. High-Resistance LC Lens has several advantages, such as 2D/3D switchable function, low driving voltage, simple structure and driving circuit, and acceptable crosstalk. For software solution, digital crosstalk reduction (DCR) method had been proposed to suppress ghost image in stereoscopic 3D display. The crosstalk was reduced to less than 5%, and the acceptable viewing range was extended to sixty degree in horizontal direction on pattern retarder 3D display. For un-normal viewing stereoscopic 3D display and auto-stereoscopic 3D display, the crosstalk will become higher than 20%. To further reduce crosstalk and maintain the luminance, multi-zone digital crosstalk reduction (MZ-DCR) method was proposed to remedy and improve the image quality. According to the results of perception experimentation, different visual depth (disparity) and motion blur can also affect the perception of 3D image quality. Therefore, considering the effects of crosstalk, disparity, and motion blur, we propose an evaluation index to quantify 3D image performance for stereoscopic displays. According to human vision experimentation, the 3D image index (∆X3D) is validated to have a high positive correlation (0.8) with human perception. After combining the human visual mechanism and ∆X3D, the proposed index can represent the visual quality of static and dynamic images on stereoscopic displays.