| 摘要: | 本論文提出一種用於遠距偵查應用之單中心多尺度超大廣角光學系統之完整設計、分析與驗證方法。該系統以單中心（single-center）光學架構為核心，結合多尺度子鏡頭拼接策略，在有限系統體積下同時實現超大視角覆蓋與遠距目標解析能力。於本研究所設計之系統中，可達成水平視角 110 度、垂直視角 72 度之廣角成像範圍，並透過多子鏡頭整合拼接，實現總畫素數約 49.8 億畫素之高解析成像效能，適用於長距離目標之偵測（detection）、辨識（recognition）與鑑別（identification）等遠距偵查任務。
在光學設計階段，本研究使用 CODE V 進行單中心光學系統之成像設計與最佳化。藉由單中心光學系統中各光學面共用球心之幾何特性，使系統於理論上具備良好的像差對稱性與離軸延展能力，從而降低超大視角與遠距高解析需求並存時之設計複雜度，並有利於多尺度子鏡頭之系統化配置。
在系統效能評估方面，本研究以 Johnson Criteria 作為任務導向之核心設計依據，將遠距偵查需求直接轉化為系統解析度、視角配置與像素取樣密度等設計指標。此外，針對所提出之單中心多尺度光學系統，進行完整之光學公差分析，並考量切向與徑向方向光線對成像品質影響機制之差異，分別評估加工公差與組裝公差對系統解析效能之影響。透過蒙地卡羅分析與補償策略設定，在累積機率 97.7% 的條件下評估系統成像品質，並以調制傳遞函數（MTF）作為量化指標，驗證系統於合理公差範圍內仍可滿足遠距偵查任務之解析需求。
此外，為驗證系統於實際應用情境下之整體表現，本研究進一步使用 LightTools 進行影像級光線追跡模擬，以模擬真實遠距偵查條件。透過此影像模擬流程，可同時評估多尺度子鏡頭於拼接前後之影像連續性，以及雜散光與鬼影效應對整體影像品質之影響，從系統層級驗證所提出架構於遠距偵查應用上的可行性。
;This dissertation proposes a complete design, analysis, and validation methodology for a monocentric ultra-high-pixel wide-angle high-resolution optical system for long-range reconnaissance applications. The proposed system is based on a monocentric (single-center) optical architecture and integrates a multi–sub-lens image stitching strategy. Under an ultra-wide field-of-view condition of 110° horizontally and 72° vertically, the system simultaneously achieves a total resolution of approximately 4.98 billion pixels (5 billion pixels), and is suitable for long-range reconnaissance tasks such as target detection, recognition, and identification.
In the optical design stage, CODE V is employed to perform image-forming optical design and optimization of the lens system. By utilizing the geometric characteristic of a monocentric optical system, in which all optical surfaces share a common center of curvature, the system theoretically exhibits good aberration symmetry and off-axis extension capability. This characteristic effectively reduces the design complexity associated with simultaneously achieving large field of view and high resolution.
For system performance evaluation, Johnson Criteria is adopted as the mission-oriented core design basis, enabling long-range reconnaissance requirements to be directly translated into design parameters including system resolution, field-of-view configuration, and pixel sampling density. In addition, a comprehensive optical tolerance analysis is conducted for the proposed monocentric multi–sub-lens optical system. The differences in image degradation mechanisms between tangential and sagittal ray directions are explicitly considered, and the impacts of manufacturing tolerances and assembly tolerances on system resolution performance are evaluated separately.
Monte Carlo analysis combined with compensation strategies is performed to assess system image quality under a cumulative probability of 97.7%, and the modulation transfer function (MTF) is used as a quantitative metric to verify that the system can maintain resolution performance that satisfies reconnaissance mission requirements within reasonable tolerance ranges.
Furthermore, to validate the overall system performance under practical application scenarios, LightTools is employed to conduct image-level ray-tracing simulations that emulate realistic long-range reconnaissance conditions. Through this image-based simulation process, both the image continuity before and after multi–sub-lens stitching and the effects of stray light and ghost reflections on image quality are evaluated. |
