具有次圖素準確性的高解析度平面定位系統之研究

Abstract

在自動化工業和電腦視覺的應用上，用高精密度視訊設備來作測量是不可或缺的步驟。其中必須以最大的正確率，來定位數位影像中測試圖樣的精密位置，以判斷測試圖樣究竟做了何種方式的移動。在這篇論文中所提出的定位系統，可以求得測試圖樣中特徵點的精密位置。這個方法包含了三個主要的步驟:前處理、參考點建立、以次圖素演算法計算特徵點的位置。在前處理中，用統計的技巧對影像序列做處理以求得標準影像。接著建立參考座標系且對參考點做編號。最後，利用參考點所在的座標，使用抗雜訊能力較佳且解析度更高的模糊次圖素運算子，求得圖樣中的邊緣方程式，並計算出特徵點的精密座標。由實驗結果顯示，此系統有不錯的估測的效果，並可應用於當CCD感測器被裝配在攝影機模組時，有效的求得CCD感測器的精密位置。

The calibration of a camera in automation industry and computer vision is an indispensable step when high precision vision is used as measuring instrument. To do this, one needs to localize the precise position with a maximum of accuracy for determining the movement of the testing pattern. A calibration system proposed in this thesis can support to obtain precise feature points of the testing pattern in the digital image. There are three major stages for this system: image preprocessing, reference points establishment, feature points extraction by using subpixel algorithm. In the preprocessing stage, some statistical techniques are used to obtain the incoming image from image sequence. Next, the reference coordinate is set up and the reference points are numbered. Finally, the proposed fuzzy subpixel operator which resists noise better and provides more precision resolution is used to extract the edge equation, then the location of feature points is estimated with subpixel accuracy. Experimental results are provided to demonstrate proposed scheme with good calibration rate, and the method can also be applied to estimate the precise position of the CCD sensor which is set up into the camera module correctly. 1.1 Surey 1.2 Motivation 1.3 Organization of the Thesis 2 Image Processing and Subpixel Precision Algorithm 2.1 Image Processing 2.1.1 Image Thresholding 2.1.2 Hough Transform 2.1.3 Median Filter 2.1.4 Run-Length Coding 2.2 Subpixel Precision Algorithm 2.2.1 One-Dimensional Edge Operator 2.2.2 Extension to Two-Dimensional Edge Patterns 2.3 Fuzzy Reasoning 3 High-Precision Technology Based on Subpixel Precision for Planar Calibration System 3.1 Statistical Preprocessing 3.2 Image Preprocessing 3.3 Reference Coordinate Establishment and Subpixel Equation Quotation 3.3.1 Reference Coordinate Establishment 3.3.2 Subpixel Equation Quotation 3.4 Information Extraction 3.4.1 One-Dimensional Fuzzy subpixel operator 3.4.2 Extension to Two-Dimensional Subpixel Operator 3.4.3 Estimation Subpixel Feature points 3.4.4 Extension of Subpixel Operator and Resistance of noise 4 Experimental Results 4.1 Experimental Environment 4.2 Experimental Results 5 Conclusions