臺灣博碩士論文加值系統

小波轉換是最近幾年所發展出來的理論，而小波轉換適用於區域性影像特性或短暫的訊號分析，其不但擁有頻率域的特性，也擁有時間域上的區域特性。因其具有多重解析度的特性，所以可以利用小波轉換來分析結構性紋路上的缺陷。
在一般軟性印刷電路板上紋路構成的方向，分為橫向、直向，和45度或135度的斜向紋路為主；檢測主要分為兩個部分來進行：針對於橫向和直向部份，原始影像做小波轉換偵測缺陷位置；針對斜向部份，對於原本的影像做擴展的步驟，使其原本的影像擴展成為一個新的菱形，並且使用斜向小波轉換來偵測缺陷。所以對於所偵測原始影像，不論紋路所構成的方向為何，都不用經過將原始影像經過旋轉的步驟。
在擷取影像的部份，我們使用硬體機台跟CCD來擷取影像。在擷取影像我們分為兩部份：首先解決水波紋對影像的影響，利用程式來降低CCD的掃描頻率；另一部分是要解決掃描平台有雜質跟線路有不平滑的邊緣。
本篇論文主要目的為提出使用小波轉換和斜向處理之小波轉換（Diagonalwise Wavelet Transform）及結合硬體機台來實現自動化缺陷檢測，並且結合CCD擷取影像的部份，使其達到軟/硬體結合，達到真正自動化缺陷檢測的目的。

Wavelet transform is the new theory that has been developed in these years, and wavelet transform applied to frequency and time domain characteristics. Due to wavelet transformation has multi-resolution analytical characteristic, which can use wavelet transform to analyze defects of structural texture possible.
The FPCB board is constructed from three structural textures: vertical, horizontal or oblique in degrees of 45 or 135 diagonal. In the method, the inspection is mainly divided into two parts. First part mainly aims at defects inspection of the horizontal and vertical circuit lines. The part makes wavelet transform in accordance with the original image, and reconstructions of the defect parts existing. The other part will spread the original image to become the new rhombus, and use the DWT transform to detect the defects. So to detect the original images, that can’t turn the images.
In the images input scheme, we integrate CCD camera, robot arm and related hardware to catch the images of FPCB. There are two parts of image reprocessing, one to solve the ripper effect by slow down the CCD camera scanning speed; the other is to solve the scanning table with the impurity by smoothing the image.
This thesis aims mainly at the defects inspection of the oblique circuit lines on FPCB to lower the operation cost and procession time, and improve the efficiency accuracy and realization possibility of the hardware to reach the goal of automatic defects inspection.

ABSTRACT (in Chinese) Ⅰ
ABSTRACT (in English) Ⅱ
ACKNOWLEDGEMENT Ⅲ
CONTENTS Ⅳ
LIST OF FIGURES Ⅵ
LIST OF TABLES Ⅷ
CHAPTER 1 INTRDUCTION 1
1.1 Introduction 1
1.2 Research Motivation 2
1.3 The organization of this thesis 2
CHAPTER 2 FPCB INSPECTIONS 3
2.1 Background on Digital Image Processing 3
2.2 Flexible Printed Circuit Board 5
CHAPTER 3 WAVELET TRANSFORM 9
3.1 Introduction to Wavelets 9
3.2 Haar Scaling Function 14
3.3 Haar Discrete Wavelet Transform…………………………………………….17
3.4 Diagonalwise DWT with Data Reflection …………………...21
CHAPTER 4 SYSTEM INTERACTION 30
4.1 Inspect System Model……………………………........…...............................30
4.2 Inspect System Interaction 31
4.3 Controlling Interface 35
4.4 Inspect System Interaction 36
4.4.1 Parameters Instruction 36
4.4.2 Image Catching Interface 37
4.5 Image Catching Problems and Strategy 38
CHAPTER 5 EXPERIMENT RESULTS 41
CHAPTER 6 CONCLUSIONS 49
REFERENCES 51

[1] Haralick, R. M., K. Shanmugam, and I. Dinstein, "Textural features for image
Classification," IEEE Transaction on Systems, Man, and Cybernetics, Vol. 3, no. 6,
pp.610-621, 1973.
[2] T. C. Li, "Defect Detection for Flexible PCB via Wavelet Transform," Master Thesis, Institute of Communication Engineering Tatung University, July 2004.
[3] Y.J.Teng, "Defect Detection for Flexible PCB by using Diagonalwise Wavelet Transform," Master Thesis, Institute of Communication Engineering Tatung University, July 2005.
[4] Rafael, C. G. and E. W. Richard, Digital Image Processing. Addison Wesley, 1992.
[5] Burrus, C. S., R. A. Gopinath and H. Guo, Introduction to Wavelets and Wavelet
Transforms A Primer, Prentice Hall, 1998.
[6] Lemarie, P. G.. and Y. Meyer, "Ondeletters et bases Hilbertiennes", Rev. Mater.
Ibero Am. Vol.2, pp.1-18, 1986.
[7] Mallat, S. G.., "A theory for multi-resolution signal decomposition: the wavelet
representation", IEEE Transactions on Pattern Analysis and Machine Intelligence,
Vol. 11, pp. 674-693. 1989.
[8] Lambert, G.. and F. Bock, "Wavelet method for texture defect detection," IEEE Int.
Conf. Image Process., Santa Barbara, CA, Vol. 3, pp. 201-204, 1997.
[9] Amet, A. L., A. Ertuzun and A. Ercil, "Texture defect detection using sub-band
domain co-occurrence matrixes, " Image Anal. Interpretation, Vol. 1, pp. 205-210, 1998.