三維外形量測之非線性數位條紋誤差校正

Abstract

本文主要研究在數位條紋投射法中，因LCD投影機與數位相機所產生的非線性條紋誤差，並提出誤差校正策略。首先建立一個強度查詢表，用來儲存電腦中設計的理想強度與實際量測強度之間的差異，並根據此差異建立另一個相位誤差查詢表，應用於相位誤差補償。接著將強度校正與相位誤差補償，應用於不同色彩與材質的待測物，進行初步的三維物體外形量測，並比較兩種方法的優缺點。其中，使用相位誤差補償能較有效消除輪廓表面的週期條紋雜訊，而強度校正能局部改善輪廓斷裂的情形。最後結合強度校正與相位誤差補償的優點，提出一套誤差校正策略，為使用具對比導引之強度校正，及殘留相位誤差補償。透過具有不規則曲面、不同表面色彩與材質待測物的三維外形量測結果可知，本文提出之誤差校正策略能有效改善重建輪廓的品質。

This thesis studies an error correction method to reduce nonlinear fringe errors induced by the LCD projector and digital cameras for measuring 3D shapes of objects. Firstly, an intensity look-up table is established for storing the difference between the assigned projected intensities and the measured intensities, and a phase error look-up table is also established for correcting the phase error induced by the intensity inconsistencies. Then, both intensity correction and phase error compensation are carried out for reconstruction of 3D object shapes, and the results between the uncorrected and corrected shapes are compared. It shows that the phase error compensation can effectively reduce the periodic fringe noise, and the intensity correction can improve the local splits on the reconstructed 3D shapes. Finally, by combining the advantages of the intensity correction and the phase error compensation, contrast-guided intensity correction and residual phase error compensation are proposed. 3D object shapes of several specimens with different colors and materials are measured. The results demonstrate that the proposed advanced error correction strategy offers good feasibility for improving the quality of reconstructed 3D shapes.