簡易檢索 / 詳目顯示
| 研究生： |
牛恆嶽 Yu-heng Niu |
|---|---|
| 論文名稱： |
以Total Variation為基礎結合適應性演算法抑制灰階圖形中的高斯雜訊 An method based on total variation and adaptive algorithm to minimize the Gaussian noise in grayscale graphics |
| 指導教授： |
王乃堅
Nai-Jian Wang |
| 口試委員: |
蔡超人
Chau-Ren Tsai 鍾順平 Shun-Ping Chung 呂學坤 Shyue-Kung Lu 劉昌煥 Chang-Huan Liu |
| 學位類別： |
碩士 Master |
| 系所名稱： |
電資學院 - 電機工程系 Department of Electrical Engineering |
| 論文出版年： | 2010 |
| 畢業學年度： | 98 |
| 語文別： | 中文 |
| 論文頁數： | 55 |
| 中文關鍵詞： | 尤拉-拉格郎日方程式 、Total Variation 、梯度下降法 、Adaptive TV |
| 外文關鍵詞： | Adaptive TV, Gradient Decent, Total Variation, Euler-Lagrange Equation |
| 相關次數： | 點閱：132 下載：2 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
數位信號處理是二十世紀最有影響力的技術之一，創新的技術已廣泛改變許多不同的領域。台灣目前的通信，影像醫療，聲學，高傳真音響，多媒體，及工業控制等產業，都發展出個自的數位信號處理技術。在這些產業中，數位信號的雜訊抑制，是非常重要的課題。
雜訊無可避免的存在於每個數位和類比系統之中。傳統上認為雜訊為高頻信號，若是所需的資訊存於低頻信號之中，最簡單的方式就是用低通濾波器作雜訊抑制。其中最有名即為高斯濾波器(Gaussian Filter)。而後發展其他出各式各樣的低通濾波器，皆是僅針對高頻雜訊作抑制。反觀，若要抑制存於高頻信號之中的雜訊，處理上就變的相當的困難。
本文提出結合R.O.F TV和Adaptive TV兩種演算法，運用離散餘弦轉換標記高頻信號區域，將同時存於高頻信號和低頻信號之中的雜訊做有效的抑制。我們將本篇論文所提出的演算法，和單一的R.O.F TV和Adaptive TV演算法做實驗比較。結果發現R.O.F TV結合Adaptive TV的演算法，在不同的測試條件下，都有較佳的效果。
Digital signal processing is one of the most influential technologies in the twentieth century, and its innovation has caused great changes in many different fields. The major industries in Taiwan, like communications, medical imaging, acoustics, high-fidelity audio, multimedia, and industrial control, all have developed their own digital signal processing technology. Noise reduction in Digital signal is a very important topic in these fields.
Noise inevitably presents in every digital and analog systems. Traditionally, noise is regarded as undesired high frequency signal and can use low-pass filter for noise reduction. The most famous low-pass filter is Gaussian Filter. The other variety of low-pass filters all simply focus on high frequency noise. It becomes quite difficult to eliminate noise in high frequency signals.
This thesis proposes a method which combines R.O.F TV, remark high frequency area by Discrete Cosine Transform and Adaptive TV algorithm to suppress the noise in both high frequency signal and low frequency signal. We compare the experiment data of single R.O.F TV and single Adaptive TV with R.O.F TV + Adaptive TV algorithm, and the results show our method is a better algorithm than other two in variety conditions.
[1] L. I. Rudin, S. Osher and E. Fatemi. “Nonlinear total variation based noise removal algorithms, ” Physical D, vol 60, pp259-268. (1992)
[2] C. R. Vogel and M. E. Oman. “Iterative methods for total variation denoising, ” SIAM Journal on Scientific Computing, vol 17, pp227-238. (1996)
[3] J. G. Rosen, “The gradient projection method for nonlinear programming, ” Part II, Nonlinear constrains, Journal of Society for Industrial and Applied Mathematics. 9, 514. (1961)
[4] 張智星, 孫春在, 水谷英二, Neuro-Fuzzy and Soft Computing, Prentice-Hall Inc. (1997)
[5] D. G. Luenberger , Yinyu Ye, Linear and Nonlinear Programming, Springer Science, Third Edition, pages 321-341. (2008)
[6] G. Gilboa, N. Sochen, Y.Y. Zeevi, “Texture preserving variational denoising using an adaptive fidelity term, ” Proc. Variation and Level-Set Methods, Nice, France, pages. 137-144. (2003)
