近年立體影像之相關應用隨著3D電影、撥放設備的普及而蓬勃發展，但目前因3D內容製作成本仍高居不下，所導致匱乏的3D數位內容是造成市場難以活絡的主要原因。在這樣的環境下，將2D影像即時地轉換成立體影像的方式便成為了一個關鍵技術。目前較廣為應用的2D轉3D方法是先透過深度圖的估計，再以深度影像繪圖法(Depth Image Based Rendering，DIBR)分別產生兩張或兩張以上的不同視點的立體影像。然而這些利用2D-to-3D技術轉換的立體影像在影像內容中存在著深度訊息不合理問題，於是本研究延伸學者Overington於2011年提出的以Naked Eye 3D演算法來增強2D影像之裸眼深度訊息的概念，藉此增強立體影像的感知深度。透過實驗方式，探討這些由2D-to-3D演算法合成的立體影像經過Naked Eye 3D演算法處理是否能達到深度增強之效果，實驗中將控制因子Image(15張具有不同深度訊息的立體影像)、 Repetition(影像處理重複次數，6次、12次、18次)、DoG Filter(高斯差濾波器尺寸，3×3、5×5、7×7)、Sigma(高斯函數標準差，0.50、1.00、1.50、2.00、2.50、3.00)。再以人類視覺以及對深度線索的認知能力，對經Naked Eye 3D演算法處理之立體影像進行深度訊息合理性的比較。採完全相依樣本受試者內四因子實驗設計，以李克特氏八分量表為評分工具，讓受試者對Naked Eye 3D演算法處理前後的立體影像對深度表現進行認知評估。所得數據以重複量數的單因子變異數分析，得到以下結論：各因子的水準之間在深度增強表現上有達顯著差異；在因子Sigma 2.0和2.5之間容易和因子Repetition、DoG Filter出現交互作用；除了Sigma 0.50，其餘因子、水準均對深度增強有正向影響。

Due to the stereo movies and popularization of stereo displays in recent years, the applications of stereo are growing rapidly. However, the high-cost of production makes the deficiency of stereo content, and it furthermore causes the market of stereo displays inactive. In such environment, converting 2D to 3D by real-time becomes a key technique to satisfy the needs of the market. In order to convert 2D to 3D, by estimating and creating a depth map from a 2D image, then using Depth Image Based Rendering method (DIBR) to generate two or more images to simulate different viewpoints is common and widely applied for the moment. However, the content of these 3D images, which are simulated from 2D images, are accompanied with some problems, such as incorrect depth information. In2011, Overington bring up the idea based on Naked Eye 3D algorism to enhance the perception of depth in the 2D images and this study is citing this idea to enhance the perception of depth in the images which were converted from 2D to 3D. The objective of this study is investigating if the simulated 3D images, which processed by Naked Eye 3D algorism, will perform better in visual depth perception than the images non-processed or not. The experiment of the study will use the procedure and parameters of Naked Eye 3D algorism as factors, which are Repetition (3 levels), DoG (Difference of Gaussian) Filter (4 levels) and Sigma (6 levels), then asking participants for applying the Likert 8-points scale to compare the 3D images processed by Naked Eye 3D algorism with the non-processed 3D images according to their cognitive perception of depth. The conclusions are listed as follows. First, all main factors are significantly different; Second, there is interaction effect between Repetition and Sigma, and DoG Filter and Sigma as well. Special in Sigma levels are 2.0 and 2.5. Final, all of the levels of each factors have positive effects to enhance the perception of depth.

Chao-Chung Cheng, Chung-Te Li, and Liang-Gee Chen. (2010a). A Novel 2D-to-3D Conversion System Using Edge Information. IEEE Transactions on Consumer Electronics, 56, 1739-1745.
Chao-Chung Cheng, Chung-Te Li, and Liang-Gee Chen. (2010b). Video 2-D–to–3-D conversion based on hybrid depth cueing. Journal of Society for Information Display, 18/9, 704-716. doi: 10.1889/JSID18.9.704
Charles Wheatstone, F.R.S. (1838). Contributions to the Physiology of Vision.—Part the First. On some remarkable, and hitherto unobserved, Phenomena of Binocular Vision. . from http://www.stereoscopy.com/library/wheatstone-paper1838.html
Goldstein, E. Bruce. (2002). Sensation and Perception (Sixth ed.): Wadsworth.
Goodwin, C. James. (2009). Research In Psychology: Methods and Design (6th ed.): John Wiley & Sons, Inc.
Hohenstatt, Peter. (1998). Leonardo Da Vinci: 1452 - 1519: Konemann.
Overington, Ian. (2011). Naked Eye 3D perception from conventional 2D displays.
Thurstone, L. L. (1927). A Law of Comparative Judgement. Psychological Review, 34, 273-286.
劉有書. (2010). 數位影像品質之分析與評估Analysis and Evaluation of Digital Image Quality. (Master), 國立中央大學.
劉楷哲、吳其霖、黃偉豪、陳信榮、李錕、羅豐祥. (2010). 基於3D顯示器格式之即時3D內容合成技術. Images & Recognition 影像與識別, 16, 76-95.
賴文能、陳韋志. (2010). 淺談 2D 至 3D 視訊轉換技術
Images & Recognition 影像與識別, 16, 61-75.