基於紋理分析與模糊理論對白光干涉檢測系統之研究

Abstract

在眾多高科技產業中，諸如半導體、平面顯示器、光纖通訊、微機電、生物醫學與電子封裝等，由於微結構表面輪廓的準確性決定了產品的效能與功能，在其製程中皆需針對微結構的表面輪廓品質進行監測。由於大部分白光干涉三維檢測系統的演算法，大多只考慮定點像素在每張影像的關係，也就是只考慮到垂直方向，如此作法並無完善地利用白光干涉產生的紋理特徵。本論文所討論的白光干涉三維檢測系統演算法，考慮到利用白光干涉產生的區域性紋理特徵，使用紋理分析、影像處理以及模糊理論，可濾除無干涉資料點的影響，並減少雜訊造成的錯誤高度數據。垂直解析能力最高達到0.1μm，量測重複度可達到2μm，並具備多種檢測功能，對於斷差高度、夾角、面積、體積、粗度、起伏以及薄膜厚度等需求，提供一個精確的檢測方案。

Among the many high-tech industries, such as semiconductors, flat panel displays, optical communications, MEMS, biomedical and electronic packaging, the micro-structure of the surface contour determines the accuracy of product performance and function in its manufacturing process units are required for the micro- structure of the surface contour to conduct quality monitoring. As most of the three-dimensional white light interferometer detection system algorithms, most only consider the fixed relationship between the pixels in each image, that is, taking into account only the vertical direction, so there is no perfect approach to the use of white light interference of the texture features. Discussed in this paper three-dimensional white light interferometer detection system algorithms, taking into account the use of white light interference of regional texture features, the use of texture analysis, image processing and fuzzy theory can filter the data points in addition to non-interference effects, and reduce errors caused by noise height data. Capability of vertical resolution up to 0.1μm, measured repeatedly degrees up to 2μm, and have a variety of detection for the off poor height, angle, area, volume, coarse degree, ups and downs as well as film thickness and other requirements, to provide a sperm does the testing program .