臺灣博碩士論文加值系統

雜訊降低在影像處理中是一個重要的環節，一個良好的影像處理演算法可以將雜訊有效的降低並且保留影像的細節部份。

在本篇論文中，首先我們介紹數種基本的雜訊型態，以及傳統的雜訊降低演算法。接著我們根據雙邊濾波器的概念改良並創造了亮度函數以及幾何函數，並以實驗顯示我們的方法對於鹽與胡椒的雜訊更加有效。同時，也顯示我們的方法比高斯雙邊濾波器更為快速。

Noise reduction is an important block in the image pipeline. Noticing noise in an image is unpleasant. A good noise reduction method can reduce the noise level and preserve the detail of the image.

In this paper, we introduce some basic noise types and traditional noise reduction methods. Then we create photometric functions and geometric functions based on the concept of the bilateral filter. We use experiments to show our proposed methods are more robust to salt-and-pepper noise. Besides, we show that our methods take less time compared with Gaussian bilateral filter.

Chapter 1 Introduction 1
1.1 Motivation 1
1.2 Introduction to Noise Types 2
1.2.1 Shot Noise (Photon Noise) 2
1.2.2 Dark Current Noise 3
1.2.3 Photo Response Non-uniformity 5
1.2.4 Read Out Noise 5
1.2.5 ADC Noise 5
1.2.6 Blooming Noise (Saturation) 7
1.2.7 Fixed Pattern Noise 8
Chapter 2 Analysis of Noise 9
2.1 Noise Level Measurement 9
2.2 Noise Models 9
2.3 Results of Previous Methods 11
2.3.1 Box Filter 12
2.3.2 Median Filter 12
2.3.3 Outlier Removal 13
2.3.4 Contrast-Dependent Outlier Removal 14
2.3.5 Smooth Replacement 14
2.3.6 K-Nearest Neighborhood 15
2.3.7 Max Filter 15
2.3.8 Min Filter 16
2.3.9 Midpoint Filter 16
2.3.10 Alpha-Trimmed Filter 16
Chapter 3 Original Bilateral Filter 18
3.1 Introduction to Bilateral Filter 18
3.2 The Gaussian Case Example 21
3.3 Disadvantages of the Tomasi’s Method 22
Chapter 4 Our Proposed Method 24
4.1 Index Definition in a Mask 24
4.2 Geometric Function 25
4.3 Photometric Function 26
4.3.1 The Difference Summation Function 26
4.3.2 The Single Difference Function 28
4.3.3 The Hybrid Function 30
4.4 The Speed-up Method 33
4.5 Other Settings 34
4.5.1 Central Weighting 35
4.5.2 Exception Prevention 35
Chapter 5 Experiments and Results 37
5.1 Test Conditions 37
5.2 Traditional Methods 40
5.3 Original Bilateral Filter and Our Proposed Bilateral Filter 41
5.4 Results for Each Noise Type 44
5.5 Comparison of Our Method and Bilateral Filter 54
5.6 Summary 61
Chapter 6 Conclusion and Future Work 63
6.1 Conclusion 63
6.2 Future Work 64

[1]M. C. Gino, “Noise, Noise, Noise,” http://www.astrophys-assist.com/educate/noise/noise.htm, 2002.
[2]R. C. Gonzalez and R. E. Woods, Digital Image Processing, 2nd Ed., Prentice-Hall, Upper Saddle River, NJ, 2002.
[3]R. M. Haralick and L. G. Shapiro, Computer and Robot Vision, Vol. I, Addison Wesley, Reading, MA, 1992.
[4]T. Z. Regier, “Development of a Gamma-Ray Beam Profile Monitor for the High-Intensity Gamma-Ray Source,” Master Thesis, Department of Physics and Engineering Physics, University of Saskatchewan, Saskatoon, 2003.
[5]C. Tomasi, R. Manduchi, "Bilateral Filtering for Gray and Color Images," Proceedings of International Conference on Computer Vision, Bombay, pp. 839-846, 1998.
[6]J. Wilkinson, “Production and Transmission of Light,” http://www.life.umd.edu/faculty/wilkinson/BSCI338/L8ltproduct/, 2002.