Research On The Parallel Astigmatic Confocal Detecting System And Its Information Processing Technology For Measurement The3-D Profile Of Micro-structure

The technology of the parallel confocal detection has played an important role in the measurement the3-D profile of the micro-structure due to its good features, e.g. non-contact measurement, high accuracy and high resolution,3-D sectioning strength, rapid detection as well as easy to implement3-D visualization measurement. In recent years, reports indicate that various types of parallel confocal detecting systems which can execute the non-scanning confocal detecting have been emerging. But most of them have been restricted to inspection biological slices. The parallel confocal detecting systems which can be used to detecting the3-D profile of micro-structure still have been theoretical exploratory. There are still great gaps between the existing measurement accuracy for the3-D profile of micro-structure and claimed. And the corresponding commercial instruments still has not come out.The title of this dissertation is "Research on the Parallel Astigmatic Confocal Detecting System and its Information Processing Technology for Measurement the3-D Profile of Micro-structure", and comes from the People’s Republic of China natural science fund project (Fundamental Research on Measurement the3D Profile of Micro-structure Based on the Detecting Principle of Parallel Astigmatic Confocal,50775063). The main points of the dissertation includes modeling on the3-D diffraction coherent imaging of the parallel confocal detecting system based on the astigmatic principle, building the experimental system and with which verifying the theoretical research results, researching the measured profile’s3-D information extraction methods and by which the performances of the system are estimated, researching visualization method to measure the profile of microstructure based on3-D reconstruction, et al. The research work is significant and substantial in the theory and application to promote the practical application of parallel confocal detecting technologies.The main work and the originalities of the dissertation are as follows:1. According to3-D diffraction coherent imaging theory, the3-D diffraction coherent imaging mathematical models of the parallel confocal detecting system based on the astigmatic principle are derived. The dissertation proposes that the essential light path of the parallel confocal detecting system based on the astigmatic principle should conform to the telecentric light path, and proves that increasing system’s magnification, selecting two exact uniform cylindrical thin lenses to form the astigmatic unit, optimizing the unit’s parameters are good for raising the whole field detecting curves’ sensitivity, symmetry, consistency and stability. Through theoretical analysis and simulation, the dissertation points out that a common optical limitation should not be set in the parallel confocal detecting system based on the astigmatic principle, otherwise the slope of the measured surface could disturb seriously astigmatic effect and result in obvious distortion of the detecting curves.2. In view of usability and reliability, the system’s magnifying power is set20. Through the Zemax simulation, the system’s transverse and axial performances are investigated and the system’s structural parameters are determined. It is also proved that optimizing the distance between two cylindrical lenses can optimized the sensitivity of whole field detecting curves.3. Several primary algorithms that deal with extracting whole field parallel probing points’ in-focus information have been designed and implemented. An effective in-focus information extraction strategy, first to create a data template and than to calculate whole field four-quadrant differential light statistics based on the template, has been proposed and executed. Two algorithms, i.e."fixed-point extracting" and "floating-point extracting" are proposed in order to remove the effect coming from parallel faculae’s transverse drift and ensure whole field parallel probing points’ in-focus position identification accuracy.4. Several experimental studies have been executed. The experiments prove that when the distance between the two cylindrical lenses is3.4mm, the average sensitivity of whole field detecting curves can reach0.292/μm, linearity range has been±2μm. Several image preprocessing methods are investigated to improve whole field detecting curves. The experiments also prove that "darkening" images can raise the sensitivity of detecting curves’100%and "nonlinear strengthening" images can decrease17%distinguishing errors of whole field probing points’ in-focus position. By iterative measurements, the stability of whole field detecting curves is estimated, The standard deviations of the whole field four-quadrant differential light statistic are within0.0030-0.0044, which implies that the identification accuracy of whole field probing points’in-focus position can reach50nm.5. Taking the uncomplicated calculation and reconstruction accuracy into account, bi-cubic-spline interpolation curve surface fitting method is selected to reconstruct the measured profile of micro-structure. A model about the error between the triangular plane and local bi-cubic-spline interpolation curve surface is derived, which can be used to not only assess reconstruction accuracy for measured micro-structure but also estimate necessary subdivision rate to a specific micro-structure according to a given reconstruction accuracy. Two mouse pickup modes,"the point pickup" and "the feature pickup", are designed to make the system adapt easily different measurement tasks.The research results could have provided some valuable design ideas and methods for commercializing the parallel confocal detecting system. The author has accomplished several significant explorations within improving the performance, raising the resolving power and measurement accuracy as will as strengthening adaptability of the parallel confocal detecting system, and accumulated some rewarding experiences.