基於景深辨析與計算之演算法研究

在光學成像的領域中，景深是一個常見用以辨識影像品質好壞的指標。拍攝時的焦距，光源位置，鏡頭的光圈大小等都會直接影響景深。在一般攝影的情況下，人為因素所造成的影像失焦問題是很常見的。本論文提出一種利用梯度單位變化量之演算法針對影像景深的問題進行辨識以及計算。在分析影像階段，在幾張影像當中選取一張品質最佳之影像，利用B-spline影像對位演算法將人為因素所產生之影像位移進行修正，並利用成像梯度演算法計算對位後之影像，計算欲觀察之理想區域面積的梯度單位面積變化量以及評估該影像之景深大小，並以變化量最大之影像作為最後選定之影像。本研究以模擬出之影像景深與實際拍攝之影像景深相比，可明顯發現，在梯度變化之影像辨識過程中之誤差值小於0.01mm，並將此技術應用於各式相機鏡頭以及顯微鏡成像中，更為準確地計算任一影像之景深。期望日後能將此演算法應用於手機通訊上，並提供使用者在選擇影像品質的過程中一個更優化的成像辨識門檻。

關鍵字

景深；梯度單位面積變化量

並列摘要

In the field of optical imaging, depth of field (DOF) height is a common indicator utilizing identify image quality. Focal length of shooting, light source position and the size of lens aperture, etc., directly affect the depth of field. In the case of general photography, it is common that images out of focus caused by human factors. This thesis proposes an algorithm based on the identification and calculation of image depth field utilizing variation of unit gradient. In the stage of image analysis, utilizing B-spline algorithm to fix image displacement caused by human factors and select an image with best quality among several images. Utilizing algorithm of image gradient to computed the image after registration and assessed the amount of depth field in a ROI and selected an image with the maximum variation of unit gradient as the final selection. This thesis compared the image DOF in simulation with actual shooting. Standard deviation of the gradient variation in the image recognition process obviously found that less than 0.01mm. In order to apply this technique in all kinds camera lens and microscope to accurate image DOF with a better accuracy. This algorithm can be applied on mobile communication and offered users a better image recognition threshold in the selecting process with expectation.

並列關鍵字

DOF；ROI；Variation of Unit Area Gradient

參考文獻

Transactions on Pattern Analysis and Machine Intelligence, Vol. 4, No. 8,

[2] V. Tympel, Carol J. Cogswell, Tony Wilson, “Three dimensional animation with a conventional light microscopy”, SPIE, Vol.2984, pp.190-198, 1997.

[3] Adams, A.,“The Camera,” New York Graphic Society, Boston, 1980.

[4] A. Castorina, A. Capra, S. Curti, M. La Cascia, V. Lo Verde, “Extension of the depth of field using multi-focus input images”, IEEE International Symposium, Vol.1-3, pp.146-150, 2004.

[5] Shutao Li, James Tin-Yan Kwok, Ivor Wai-Huang Tasng, and Taonan Wan, “Fusing Images With Different Focuses Using Support Vector Machines”, IEEE Transactions on neural networks, Vol.15, No.6, Nov 2004.

國際替代計量

基於景深辨析與計算之演算法研究

未授權

主題瀏覽