3D電腦生成全像系統之建構與影像顯示及評估方法

Abstract

本論文主要的研究目標在於探究電腦生成全像的優化顯示方法，並從演算法、影像評估方法、光學機構三方面進行優化與設計。 基於優化型 Gerchberg-Saxton 演算法(Modified Gerchberg-Saxton Algorithm, MGSA)的架構下，本研究提出以空間多工的方式，藉由避除零階光與共軛光，使影像達到優化效果。此外，本研究針對觀看視野範圍，提出有效視域演算法(Effective Viewshed Algorithm, EVA)，並以此方法進行資訊點範圍的選取，實現於觀看視野範圍內的影像資訊最佳化重建。另外，對於多影像的同軸顯示，提出相位相位加密(Phase-locked)的方式，有效降低了非解像資訊的干擾。基於MGSA的架構，本研究提出之功能擴充，都具備顯示優化之結果。 另一方面，大多數的影像評估方法都是為二維資訊所設計，因為三維電腦全像包含多深度之資訊點，一般的評價方法會因為未解像資訊而影響評估結果。本研究率先提出了三維結構相似性指標(3D structural similarity index, 3D SSIM)的評估方法，基於SSIM的公式進行修正，並提出對應評估系統。所提之3D SSIM的評估方法，有效解決二維評估方法對於全像影像的失真估算。該方法對於三維電腦生成全像而言，將會是往後重要的評估指標與方法。 最後，針對全像色彩再現系統進行優化設計與討論。在系統設計的過程中，對於不同波長間的影像大小與重建位移誤差，以演算法空間多工的方式成功完成影像資訊的修正。並以單一波導的光學設計方法，改善合光系統的複雜度與面板使用率，並將整體系統大小縮小86.6%。並因應雷射光斑問題，提出震動光波導之方法，將重建影像之光強度被均化，使雷射光斑受到了抑制，以此改善了影像品質。

In this dissertation, the goal is to investigate the optimized display method for computer-generated holograms, as well as to optimize algorithms, methods for evaluating images, and optical systems. Using the Modified Gerchberg-Saxton Algorithm (MGSA) framework, this work proposes an image optimization technique that avoids zero-order light and conjugate light. Further, we propose the Effective Viewshed Algorithm (EVA) for selecting the range of information points within the viewing field to optimize the reconstruction of the image information. In addition, a phase-locked approach is proposed for displaying multiple images concurrently, which effectively reduces crosstalk caused by non-reconstructed information. According to the MGSA, the functional extensions proposed in this study have optimized results. The experimental image evaluation methods are generally designed for two-dimensional information. In 3D holograms, the non-reconstructed information will affect the general evaluation result. This study is the first to propose a 3-D SSIM evaluation method based on the SSIM formula and a corresponding evaluation system. In comparison to the 2D SSIM method, the 3D SSIM evaluation method effectively estimates quality of holographic images. In the future, this method will be an important evaluation indicator and method for 3D computer-generated holograms For the full-color holographic display system, the optimization method is designed and discussed. Using the algorithmic spatial multiplexing method, it is possible to successfully correct the image size and reconstruction displacement errors between different wavelengths within the system design. Single waveguide optical design reduces the overall system size by 86.6% and improves the complexity of the optical system. Furthermore, we propose a method of vibrating waveguides to resolve the speckle problem, in which the intensity of the reconstructed image is homogenized to reduce speckles and you get a better image.