| 摘要: | 本論文是利用結合適應性光學和二元化散射斑紋定理來量測金屬的表面粗糙度。本論文研究重點在(1)利用二元化散射斑紋定理分析，來量測金屬的表面粗糙度，(2)探討適應性光學在動態擾流和不同折射率下所造成不同的像差與校正結果，以及(3)利用適應性光學校正光源像差並結合粗糙度光學測量。 (1) 利用二元化散射斑紋分析，量測金屬的表面粗糙度: 本實驗提出了一種新的光學方法可應用於線上表面粗糙度測量。藉由雷射照射試片表面可得到反射的散射斑紋圖案。進而將散射斑紋轉化為二元化圖案加以分析。二元化的強度分佈圖像是結合散斑和散射現象的綜合效應。本實驗建立一個新的圖像光強分佈參數值(SdBD)，並提出與機械研磨加工金屬的表面粗糙度參數 Ra的相關實驗。測量結果證明有SdBD與 Ra之間有良好的相關性，相關係數為 0.9706。在該方法的實用性，可應用在線上粗糙度測量。本研究於6個金屬研磨試片的粗糙度Ra = 0.2?6.25μm（0.3λ到10λ，其中λ為二極體雷射波長）進行實驗。 (2) 探討適應性光學在動態擾流和不同折射率界面下所造成不同的像差與校正結果： 本實驗可證明適應性光學(AO)影像系統有能力校正在動態擾流和不同折射率下所產生的像差。其中以可調變聚焦鏡(DM)用來校正畸變的波前，而Shack-Hartmann波前感測器則作為量測波前像差的感測器。在無AO校正時，我們引進熱對流使得波前RMS (root-mean-square) 上升至0.26 μm。經過AO 校正後，RMS校正為0.07 μm。此外AO 系統可在3秒內完成補償像差的工作。我們不只補償對熱對流所產生的像差，並且也消除了本身光學系統的像差。此外我們放入不同的折射率的多層的convex/concave 的界面。對於水和玻璃界面，AO系統可將RMS =2.17μm 下降至RMS=0.17μm。對於油和玻璃界面，經AO系統校正後，將RMS =0.24μm 下降至0.10μm。 (3) 利用適應性光學校正光源像差並結合粗糙度光學測量： 本研究提出一個結合適應性光學（AO）和二元化圖像分析的粗糙度光學測量系統。其主要目的是展現當線上量測產生熱擾流與流體流動，利用適應性光學校正光源像差的必要性。藉由雷射照射試片表面可得到反射的散射斑紋圖案。而將散射斑紋轉化為二元化圖案與表面粗糙度加以分析。在無適應性光學校正系統中，當引進熱擾流使得波前RMS從 0.14μm上升至1.4μm。經過AO即時閉迴路校正之後，我們可以改善波前RMS為 0.12μm。此外也對於 AO控制系統中的各種增益值進行研究，並且找到臨界增益值是能夠穩定補償波前誤差並且校正時間小於 2秒。本研究進行實驗於5個金屬研磨試片的粗糙度Ra 0.2?3.125μm（0.3λ到5λ，其中λ為二極體雷射波長），測量結果證明SdBD與 Ra之間有良好的相關性，相關係數為 0.9982。 在散射定理方面，我們也結合了適應性光學測量系統。在無適應性光學校正系統中，當引進壓縮空氣使得波前RMS上升至1.5μm。經過AO即時閉迴路校正之後，我們可以改善波前RMS為 0.14μm。。本研究進行實驗於5個金屬研磨試片的粗糙度Ra 0.2?3.125μm（0.3λ到5λ，其中λ為二極體雷射波長），測量結果證明有良好的相關性，相關係數為 0.9967。並且比較stylus 量測法，可發現其量測誤差可控制在8.7% 。因此，此系統可作為快速估計粗糙度的量測器，可以進一步提高製造過程中的工件精度和穩定性。 The aim of this study is using a newly-developed method of intensity distribution of binary image (SdBD) for measuring surface roughness under dynamic turbulence with adaptive optics (AO) system. The focus of the study is (1) roughness measurement based on spatial average analysis of binary speckle image, (2) adaptive optics system for correcting aberrations induced by under dynamic turbulence and combinative interfaces of refractive-index-mismatch, (3) Adaptive optics-assisted imaging system for roughness measurement of metal surface. (1) The speckle image was obtained by illuminating a laser beam and the reflected laser pattern image from a surface was binarizd and examined. The intensity distribution of binary image utilizes the combined effects of speckle and scattering phenomena. A new parameter of intensity distribution of binary image, SdBD has been proposed and the surface roughness parameter Ra of machined surfaces (ground) was correlated experimentally. Measurement results demonstrate an excellent correlation between the SdBD and Ra with correlation coefficient of 0.9706. The practicality of the proposed method In-situ roughness measurement was applied to six samples from roughness Ra 0.2 to 6.25μm (0.3 λ and 10 λ, where λ is diode laser wavelength) of steel through grinding process. (2). This study experimentally demonstrate the capability of adaptive optics (AO) correction on aberrations induced by dynamic turbulence and combinative effects of multiple layers with convex/concave interfaces as well as Refractive-Index-Mismatch (RIM). A deformable mirror (DM) was used as a wavefront corrector, the wavefront aberration was detected by a Shack-Hartmann wavefront sensor. In the absence of AO correction scheme, induced low temporal turbulences can severely degrade the residual RMS (root-mean-square) errors of 0.26 μm. After real time closed-loop AO correction, we can improve wavefront to RMS of 0.07 μm, which not only compensate aberration error from induced disturbances, but also overall optical system. In addition, AO was found to be able to steadily compensate wavefront errors in less than 3 seconds. For Refractive-Index-Mismatch, we consider the aberration introduced by interfaces of RIM between water/oil and glass. After adaptive optics correction, we can improve wavefront with root mean square (RMS) of 2.17 to 0.17 μm for an interface between water and glass. As for the interface between oil and glass, we are capable of improving RMS of 0.24 to 0.10 μm. The benefits of AO correction are facilitated by removing low order of Zernike modes such as defocus and tip/tilt, which are found to be the two main contributing factors in serial arrangement of convex/concave interfaces and RIM. Adaptive optics system shows correction capability for multiple layers of different geometrical interfaces with RIM. (3).This study proposes an integrated roughness measurement system based on adaptive optics (AO) and binarized analysis of speckle pattern images. The main purpose of this study is to demonstrate the necessity of AO compensation where disturbances of both heat and fluid flow exist. The speckle image was first obtained by illuminating a laser beam and the reflected laser pattern image from a surface was binarized to experimentally correlate with the intensity. In the absence of AO correction scheme, induced turbulences can severely degrade the residual RMS errors from 0.14 to 1.4 μm. After real time closed-loop AO correction, we can improve wavefront with RMS of 0.12 μm, which not only compensate aberration error from induced disturbances, but also overall optical system. In addition, AO with various gains in control system was investigated and a threshold gain value was found to be able to steadily compensate wavefront errors in less than 2 seconds. Measurement results of five steel samples from roughness Ra =0.2 to 3.125μm (0.3 λ and 5 λ, where λ is diode laser wavelength) demonstrate an excellent correlation between the SdBD and Ra with correlation coefficient of 0.9982. We have developed an in-process measurement of surface roughness by combining an optical probe of laser scattering phenomena and adaptive optics for aberration corrections. The aim of this study is to demonstrate the necessity for AO compensation in regions containing turbulence of fluid flow. In the absence of the AO correction scheme, induced turbulence can severely increase the residual root mean square (RMS) error to 1.5 μm. After a real-time closed-loop AO correction, we can reduce the wavefront RMS error to 0.14 μm. Measurement results of five steel samples having roughness ranging from 0.2 to 3.125 μm (0.3λ and 5λ, where λ is the diode laser wavelength) demonstrate an excellent correlation between the peak power and average roughness with a correlation coefficient (R2) of 0.9967. The results were verified by the stylus method using three samples of steel (AISI 304) under various mechanical grinding conditions. The proposed AO-assisted system is in good agreement with less than 8.7% error values. Therefore, the developed system can be used as a rapid in-process roughness monitor/estimator to further increase the precision and stability of manufacturing processes in-situ. |
