水下環境中的雜質經常影響水下影像的清晰程度，造成水下影像在視覺的分析及研究有很大的局限。除了雜質會造成影像模糊的情況外，光線在水中傳遞時會因為波長的不同，產生不同的折射和散射結果。水下影像的低對比度，及部分顏色衰減所形成的色偏現象，和霧氣中拍攝的影像有類似的結果。近年來，在[1]中已經提出了有效的單張影像除霧的演算法，稱為暗通道預設(Dark Channel Prior)。而在其他的許多研究中(如[2-7])則將此演算法應用在水下影像的除霧。另外在水下攝影時，經常拍攝出曝光不足的照片，這是因為相機跟人眼一樣，當同時接收順光、逆光及陰暗影像時，很難清楚判斷出物體的輪廓、細節等等。針對曝光不足的影像，Pang等人[8]提出了一個極端低對比度影像之強化演算法。此方法是基於Dong[9]的演算法所改良而成，除了可以有效的增強影像的對比度，也能夠改善Dong[9]的演算法會造成的光暈現象。因此為了進行水下的研究與分析、或改善水下監視系統這些問題，一套可以針對極端低對比度之水下影像，進行影像強化的除霧系統是不可或缺的。
本論文所提出的水下影像除霧系統包含了四大步驟：1）在RGB色彩模型下，利用直方圖分佈來判斷是否為極端對比影像或是一般對比影像，並決定後續步驟的處理方式。 2）由於水下影像和霧氣中所拍攝之影像有相似的對比度弱化、色偏等問題，因此可以利用暗通道預設技術來計算影像的大氣光光源。3）在暗通道預設技術中，傳遞率的計算尤其重要。因此在估計出影像的傳遞率圖之後，我們採用快速簡化導引濾波器來修正傳遞率圖的邊緣，避免光暈的現象過於嚴重。4）最後以除霧的方法進行影像還原，並分別對RGB三個色彩通道進行不同的吸收衰減補償。本論文以大氣光估計器為基礎，提出快速的大氣光估計器硬體架構，以縮放影像的概念，大幅減少估計大氣光所需要讀取的資料量，使得運算速度提升許多。此外，本論文以[8]的演算法為基礎，提出改良的除霧系統硬體架構，為了方便硬體實現，我們簡化其演算法，且能達到幾乎相同的效果。

The impurity in the water affects the clarity of underwater images, leading to great limitations on the application of underwater images in visual analysis and research. In addition to the fact that impurities cause image blurring, different refraction and scattering results are produced depending on the wavelength of the light, when light is transmitted in water,. The color contrast caused by the low contrast of the image and the partial color decay has similar results as the image taken in the fog. In recent years, an effective single image defogging algorithm, Dark Channel Prior, has been proposed in [1]. In many other studies (such as [2-7]), this algorithm is also applied to defogging underwater images. In addition the underexposed photos are often taken, when photographing in the underwater environment. Because the camera is like the human eye, it is not easy to clearly determine the outline, details, etc. of the object, when it receives the smooth, backlit and dark images at the same time. For the case of underexposed images, Pang et al. [8] proposed an enhanced algorithm for images with extremely low contrast. This method is based on the algorithm of Dong[9]. In addition to effectively enhancing the contrast of the image, it can also improve the halo caused by the algorithm of Dong[9]. Therefore, in order to improve these problems in underwater research analysis or underwater monitoring systems, a set of defogging systems that can perform image enhancement for underwater images with extreme contrast differences is indispensable.
The defogging system proposed in this thesis contains four major steps: 1) With the RGB color model, the histogram distribution is used to determine whether it is an extreme contrast image or a normal contrast image. Accordingly the subsequent steps have different processing methods. 2) Since the underwater image and the image captured in the fog have similar problems such as contrast reduction and color shift, the dark channel prior can be used to calculate the atmospheric light source of the image. 3) In the dark channel prior, the calculation of the transfer rate is particularly important. Therefore, after estimating the transmission rate map of the image, we use a fast and simple guided filter to correct the edge of the transmission rate map to avoid the phenomenon of halation. 4) Finally, the image is restored by the method of defogging, and different absorption attenuation compensation is performed for the three color channels of RGB. To meet the real-time requirement, this thesis proposes a fast atmospheric light estimator hardware architecture based on the subsampling technique. As the result, the amount of data that needs to be read to estimate atmospheric light is greatly reduced, and the computational speed is much improved. Moreover, based on the algorithm of [8],this thesis also proposes the improved hardware architecture of the defogging system. In order to facilitate the implementation of the hardware architecture, the algorithm is simplified and the high performance and low cost can be achieved.

論文審定書 i
論文提要 ii
摘要 iii
Abstract v
目錄 vii
圖目錄 x
表目錄 xiii
第 1 章 序論 1
1.1 研究動機 1
1.2 論文大綱 3
第 2 章 研究背景 4
2.1 大氣散射原理 4
2.2 基於暗通道預設之除霧演算法 5
2.2.1 暗通道預設 5
2.2.2 估計傳遞率 7
2.2.3 估計大氣光值 8
2.2.4 無霧影像還原 9
2.3 邊緣修正方法 10
2.4 水下影像還原方法 12
2.5 增強極端對比影像 14
2.5.1 極端低對比度之影像增強 14
2.5.2 極端低對比度之影像與有霧影像之相關性 14
第 3 章 研究方法 16
3.1 運算流程 16
3.2 極端對比度判斷 18
3.3 計算水下大氣光值 21
3.3.1 大氣光估計 21
3.3.2快速大氣光估計 21
3.4 傳遞率估計 24
3.5 傳遞率修正 25
3.6 水下影像強化 26
3.6.1水下影像還原 26
3.6.2水下影像色偏修正 26
3.7 增強極端低對比之水下影像 28
3.7.1 極端低對比之水下大氣光估計 28
3.7.2 簡化傳遞率估計 28
3.7.3 簡化傳遞率修正 29
3.7.4 簡化極端低對比之水下影像還原 29
第 4 章 提出的水下影像強化硬體架構 31
4.1 硬體架構 31
4.2 灰階轉換器[12] 33
4.3 大氣光估計器 34
4.3.1傳統大氣光估計器[18] 34
4.3.2 提出的快速大氣光估計器 35
4.4 傳遞率估計器 36
4.4.1傳統傳遞率估計器[12] 36
4.4.2 提出的簡化傳遞率估計器 37
4.5 改良之簡化快速導引濾波器 38
4.5.1傳統簡化快速導引濾波器[12] 38
4.5.2 提出的改良之簡化快速導引濾波器 38
4.6 還原模組 40
4.6.1傳統還原模組[12] 40
4.6.2提出的簡化還原模組 41
第 5 章 實驗結果 42
5.1 實驗步驟與方法 42
5.2 硬體與軟體實作比較 44
5.3 一般水下影像強化結果比較 63
5.4 硬體分析 68
第 6 章 結論與未來工作 70
6.1結論 70
6.2未來研究方向 70
參考文獻 71

[1] K. He, J. Sun, and X. Tang, “Single image haze removal using dark channel prior,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 33, no. 12, pp. 2341-2353, Dec. 2011.
[2] A. Galdran, D. Pardo, A. Picón, and A. Alvarez-Gila, “Automatic red-channel underwater image restoration,” Journal of Visual Communication and Image Representation, vol. 26, pp. 132-145, Jan. 2015.
[3] H. Y. Yang, P. Y. Chen, C. C. Huang, Y. Z. Zhuang, and Y. H. Shiau, “Low complexity underwater image enhancement based on dark channel prior,” 2011 Second International Conference on Innovations in Bio-inspired Computing and Applications (IBICA), , pp. 17-20, Dec. 2011.
[4] S. J. Horng, P. J. Liu, and J. S. Lin, “Improving the Contrast Enhancement of Oceanic Images Using Modified Dark Channel Prior,” 2016 International Symposium on Computer, Consumer and Control, pp. 801-804, July, 2016.
[5] S. Borkar, and S. V. Bonde, “Underwater image restoration using single color channel prior,” 2016 International Conference on Signal and Information Processing (IConSIP), pp. 1-4, Oct. 2016.
[6] C. Li, J. Quo, Y. Pang, S. Chen, and J. Wang, “Single underwater image restoration by blue-green channels dehazing and red channel correction,” 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 1731-1735. Mar. 2016.
[7] C. Y. Cheng, C. C. Sung, and H. H. Chang, “Underwater image restoration by red-dark channel prior and point spread function deconvolution,” 2015 IEEE International Conference on Signal and Image Processing Applications (ICSIPA), pp. 110-115, Oct. 2015.
[8] Jianhua Pang, Sheng Zhang, Wencang Bai “A novel framework for enhancement of the low lighting video”, 2017 IEEE Symposium on Computers and Communications (ISCC), pp.1-8, 2017.
[9] X. Dong, Y. Pang, and J. Wen, “Fast Efficient Algorithm for Enhancement of Low Lighting Video,” IEEE Intl. Conf. on Multimedia and Expo, pp. 1-6, 2011.
[10] Robby T. Tan,”Visibility in Bad Weather from a Single Image,“ IEEE Conf. Computer Vision and Pattern Recognition, pp.1-8, June. 2008.
[11] R. Fattal, “Single image dehazing”, Proceedings of ACM SIGGRAPH, August 2008.
[12] 張城豪, “Design and Implementation of Fast Guided Filter Hardware Architecture for Underwater Image Dehazing” 國立中山大學, 碩士論文, 2017.
[13] 影像處理實驗室http://image.cse.nsysu.edu.tw/index1.html
[14] S.M. Pizer, et al., “Adaptive Histogram Equalization and Its Variations,” Computer Vision, Graphics, and Image Processing, vol. 39, no. 3, pp. 355-368, Sep. 1987.
[15] 李尉嘉, “利用變化的直方圖同等化法及區域對比調適增強法處理低曝光影像之增強方法” 國立中興大學, 碩士論文, 2016.
[16] Yakun chang, Cheolkon Jung, Peng Ke, Hyoseob Song, and Jungmee Hwang, “Automatic Contrast-Limited Adaptive Histogram Equalization With Dual Gamma Correction,” IEEE Access, vol. 6, 2018.
[17] 鄭士弘, “極端對比差異影像之增強方法” 國立高雄應用科技大學, 碩士論文, 2013.
[18] Y.-H. Shiau, H.-Y. Yang, P.-Y. Chen, and Y.-Z. Chuang, “Hardware Implementation of a Fast and Efficient Haze Removal Method,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 23, no. 8, pp. 1369-1374, 2013.
[19] K. He, and J. Sun, “Fast guided filter,” ArXiv preprint ArXiv:1505.00996, 2015.
[20] H. Lu, Y. Li, and S. Serikawa, “Underwater image enhancement using guided trigonometric bilateral filter and fast automatic color correction,” 2013 20th IEEE International Conference on Image Processing (ICIP), pp. 3412-3416, Sept. 2013.
[21] C. Ancuti, C. O. Ancuti, and T. Haber, “Enhancing underwater images and videos by fusion,” 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 81-88, June. 2012.
[22] 郭家庭, “Underwater Image Dehazing and Enhancement by Compensating Light Absorption Loss and Multi-exposure Fusion,” 國立中山大學, 碩士論文, 2017.
[23] J. Y. Chiang, and Y. C. Chen, “Underwater image enhancement by wavelength compensation and dehazing,” IEEE Transactions on Image Processing, vol. 21, no. 4, pp. 1756-1769, April 2012.
[24] 徐立緯, “A Variable-precision Architecture for Special Function and Floating-point Multiply-add-fused Operation,” 國立中山大學, 碩士論文, 2015.
[25] Stéphane Bazeille, Isabelle Quidu, Luc Jaulin, and Jean-Philippe Malkasse,” Automatic underwater image pre-processing,” CMM’06, 2006.
[26] N. Carlevaris-Bianco, A. Mohan, and R. M. Eustice, “Initial results in underwater single image dehazing,” IEEE OCEANS 2010, pp. 1-8, Sept. 2010.
[27] J. Y. Chiang, and Y. C. Chen, “Underwater image enhancement by wavelength compensation and dehazing,” IEEE Transactions on Image Processing, vol. 21, no. 4, pp. 1756-1769, April 2012.
[28] S. Serikawa, and H. Lu, “Underwater image dehazing using joint trilateral filter,” Computers & Electrical Engineering, vol. 40, no. 1, pp. 41-50, Jan. 2014.
[29] Z. Wang, A.C.Bovik, H.R. Sheikh, and E. P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,” IEEE Transactions on Image Processing, vol. 13, no. 4, pp. 600-612, 2004.
[30] L. Zhang, L. Zhang, X. Mou, and D. Zhnag, “FSIM: A feature similarity index for image quality assessment,” IEEE Transactions on Image Processing, vol. 20, no. 8, pp.2378-2386, 2011.