請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/59175
完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.advisor | 貝蘇章 | |
dc.contributor.author | Hui-Chih Wang | en |
dc.contributor.author | 王彙智 | zh_TW |
dc.date.accessioned | 2021-06-16T09:17:13Z | - |
dc.date.available | 2017-07-20 | |
dc.date.copyright | 2017-07-20 | |
dc.date.issued | 2017 | |
dc.date.submitted | 2017-07-12 | |
dc.identifier.citation | [1] Tan, Robby T. 'Visibility in bad weather from a single image.' Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference on. IEEE, 2008.
[2] Fattal, Raanan. 'Single image dehazing.' ACM transactions on graphics (TOG) 27.3 (2008): 72. [3] He, Kaiming, Jian Sun, and Xiaoou Tang. 'Single image haze removal using dark channel prior.' IEEE transactions on pattern analysis and machine intelligence 33.12 (2011): 2341-2353. [4] Gao, Yuanyuan, et al. 'A fast image dehazing algorithm based on negative correction.' Signal Processing 103 (2014): 380-398. [5] Gibson, Kristofor B., and Truong Q. Nguyen. 'An analysis of single image defogging methods using a color ellipsoid framework.' EURASIP Journal on Image and Video Processing 2013.1 (2013): 37. [6] Zhu, Qingsong, Jiaming Mai, and Ling Shao. 'A fast single image haze removal algorithm using color attenuation prior.' IEEE Transactions on Image Processing 24.11 (2015): 3522-3533. [7] Berman, Dana, and Shai Avidan. 'Non-local image dehazing.' Proceedings of the IEEE conference on computer vision and pattern recognition. 2016. [8] Pei, Soo-Chang, and Tzu-Yen Lee. 'Nighttime haze removal using color transfer preprocessing and dark channel prior.' 2012 19th IEEE International Conference on Image Processing. IEEE, 2012. [9] Zhang, Jing, Yang Cao, and Zengfu Wang. 'Nighttime haze removal based on a new imaging model.' 2014 IEEE International Conference on Image Processing (ICIP). IEEE, 2014. [10] Li, Yu, Robby T. Tan, and Michael S. Brown. 'Nighttime Haze Removal with Glow and Multiple Light Colors.' Proceedings of the IEEE International Conference on Computer Vision. 2015. [11] Ancuti, Cosmin, et al. 'Night-time dehazing by fusion.' Image Processing (ICIP), 2016 IEEE International Conference on. IEEE, 2016 [12] Zhang, Jing, Yang Cao, and Zengfu Wang. 'Nighttime Haze Removal with Illumination Correction.' arXiv preprint arXiv:1606.01460 (2016). [13] Nayar, Shree K., and Srinivasa G. Narasimhan. 'Vision in bad weather.' Computer Vision, 1999. The Proceedings of the Seventh IEEE International Conference on. Vol. 2. IEEE, 1999. [14] Lam, Edmund Y. 'Combining gray world and retinex theory for automatic white balance in digital photography.' Consumer Electronics, 2005.(ISCE 2005). Proceedings of the Ninth International Symposium on. IEEE, 2005. [15] Liu, Fei, et al. 'Polarimetric dehazing utilizing spatial frequency segregation of images.' Applied Optics 54.27 (2015): 8116-8122. [16] Schechner, Yoav Y., Srinivasa G. Narasimhan, and Shree K. Nayar. 'Instant dehazing of images using polarization.' Computer Vision and Pattern Recognition, 2001. CVPR 2001. Proceedings of the 2001 IEEE Computer Society Conference on. Vol. 1. IEEE, 2001. [17] Shwartz, Sarit, Einav Namer, and Yoav Y. Schechner. 'Blind haze separation.' Computer Vision and Pattern Recognition, 2006 IEEE Computer Society Conference on. Vol. 2. IEEE, 2006. [18] Fattal, Raanan. 'Dehazing using color-lines.' ACM Transactions on Graphics (TOG) 34.1 (2014): 13. [19] Levin, Anat, Dani Lischinski, and Yair Weiss. 'A closed-form solution to natural image matting.' IEEE Transactions on Pattern Analysis and Machine Intelligence 30.2 (2008): 228-242. [20] He, Kaiming, Jian Sun, and Xiaoou Tang. 'Guided image filtering.' European conference on computer vision. Springer Berlin Heidelberg, 2010. [21] K. Tang, J. Yang, and J. Wang, “Investigating haze-relevant features in a learning framework for image dehazing,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), Jun. 2014, pp. 2995–3002. [22] Ancuti, Codruta Orniana, and Cosmin Ancuti. 'Single image dehazing by multi-scale fusion.' IEEE Transactions on Image Processing 22.8 (2013): 3271-3282 [23] Reinhard, Erik, et al. 'Color transfer between images.' IEEE Computer graphics and applications 21.5 (2001): 34-41. [24] Schettini, Raimondo, et al. 'Contrast image correction method.' Journal of Electronic Imaging 19.2 (2010): 023005-023005. [25] Elad, Michael. 'Retinex by two bilateral filters.' International Conference on Scale-Space Theories in Computer Vision. Springer Berlin Heidelberg, 2005. [26] Narasimhan, Srinivasa G., and Shree K. Nayar. 'Shedding light on the weather.' Computer Vision and Pattern Recognition, 2003. Proceedings. 2003 IEEE Computer Society Conference on. Vol. 1. IEEE, 2003. [27] Li, Yu, and Michael S. Brown. 'Single image layer separation using relative smoothness.' Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2014. [28] Miao, Ligang, Yanjun Chen, and Aizhong Wang. 'Video smoke detection algorithm using dark channel priori.' Control Conference (CCC), 2014 33rd Chinese. IEEE, 2014 [29] Hunt, Robert William Gainer. The reproduction of colour. John Wiley & Sons, 2005. [30] Land, Edwin H., and John J. McCann. 'Lightness and retinex theory.' Josa 61.1 (1971): 1-11. [31] Rue, Havard, and Leonhard Held. Gaussian Markov random fields: theory and applications. CRC Press, 2005. [32] Tarel, Jean-Philippe, and Nicolas Hautiere. 'Fast visibility restoration from a single color or gray level image.' Computer Vision, 2009 IEEE 12th International Conference on. IEEE, 2009. [33] Kratz, Louis, and Ko Nishino. 'Factorizing scene albedo and depth from a single foggy image.' Computer Vision, 2009 IEEE 12th International Conference on. IEEE, 2009. [34] Bahat, Yuval, and Michal Irani. 'Blind dehazing using internal patch recurrence.' Computational Photography (ICCP), 2016 IEEE International Conference on. IEEE, 2016. [35] Narasimhan, Srinivasa G., and Shree K. Nayar. 'Vision and the atmosphere.' International Journal of Computer Vision 48.3 (2002): 233-254. [36] Narasimhan, Srinivasa G., and Shree K. Nayar. 'Contrast restoration of weather degraded images.' IEEE transactions on pattern analysis and machine intelligence 25.6 (2003): 713-724. [37] Choi, Lark Kwon, Jaehee You, and Alan Conrad Bovik. 'Referenceless prediction of perceptual fog density and perceptual image defogging.' IEEE Transactions on Image Processing 24.11 (2015): 3888-3901. [38] Park, Dubok, David K. Han, and Hanseok Ko. 'Nighttime image dehazing with local atmospheric light and weighted entropy.' Image Processing (ICIP), 2016 IEEE International Conference on. IEEE, 2016. [39] Lv, Xingyong, Wenbin Chen, and I-fan Shen. 'Real-time dehazing for image and video.' Computer Graphics and Applications (PG), 2010 18th Pacific Conference on. IEEE, 2010. [40] Kim, Jin-Hwan, et al. 'Temporally x real-time video dehazing.' Image Processing (ICIP), 2012 19th IEEE International Conference on. IEEE, 2012. [41] Dong, Xuan, et al. 'Fast efficient algorithm for enhancement of low lighting video.' Multimedia and Expo (ICME), 2011 IEEE International Conference on. IEEE, 2011. [42] Zhu, Qingsong, et al. 'Mean shift-based single image dehazing with re-refined transmission map.' Systems, Man and Cybernetics (SMC), 2014 IEEE International Conference on. IEEE, 2014. [43] Xu, Haoran, et al. 'Fast image dehazing using improved dark channel prior.' Information Science and Technology (ICIST), 2012 International Conference on. IEEE, 2012. [44] Xie, Bin, Fan Guo, and Zixing Cai. 'Improved single image dehazing using dark channel prior and multi-scale Retinex.' Intelligent System Design and Engineering Application (ISDEA), 2010 International Conference on. Vol. 1. IEEE, 2010. [45] Gibson, Kristofor B., and Truong Q. Nguyen. 'On the effectiveness of the dark channel prior for single image dehazing by approximating with minimum volume ellipsoids.' Acoustics, Speech and Signal Processing (ICASSP), 2011 IEEE International Conference on. IEEE, 2011. [46] Park, Dubok, et al. 'Single image dehazing with image entropy and information fidelity.' Image Processing (ICIP), 2014 IEEE International Conference on. IEEE, 2014. [47] Zhang, Xiangdong, et al. 'Enhancement and noise reduction of very low light level images.' Pattern Recognition (ICPR), 2012 21st International Conference on. IEEE, 2012. | |
dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/59175 | - |
dc.description.abstract | 本篇論文主要探討除霧演算法在各方面之應用,包括在白天與夜間環境光源下除霧,進一步修改除霧演算法使其適用於被煙霧遮蔽之影像。論文第一個部分主要探討的是白天環境光源下的除霧演算法,基於現有的除霧演算法,dark channel prior、 color line、或是negative correction,根據這些不同的prior information,可估計出不同的transmission map。每一種prior information做出來的方法會影響最後除霧效果的表現。根據每一種prior information會對transmission map中不同的地方產生enhancement,進而達到不同的除霧效果。然而這些演算法主要適用於正常白天環境光源下的霧霾影像,在夜晚環境光源下的影像並不適用。夜霧影像有許多獨特的問題,包括人造光源以及低亮度的影響,會造成正常除霧演算法的失真,故有許多針對處理夜霧影像的方法問世。本文第二個部分主要探討夜間霧霾影像的除霧演算法,探討的方法主要有color transfer method、new imaging modeling、 glowing effect removal,以及 image fusion based 等方法,其主要演算法設計的概念為將人造光源所造成的霧霾影像的失真給去除,使得夜霧影像可以接近白天霧霾的影像,進而使用一般的除霧演算法來處理改變後的夜霧影像。第三個部分則為影像除煙。我們發現在煙霧影像中,在煙霧區域的色彩會有不均勻的失真現象,這是因為在每個color channel的粒子濃度不完全均勻分布所造成,針對此種現象,每一個color channel需要不同的transmission map來做處理。因此,我們提出single dark channel prior的觀念使用dark channel prior的除霧演算法來做影像除煙,更利用多次除霧的過程,使每一次除完的煙霧殘留更加稀少。此外,透過夜晚影像強化的演算法,強化夜晚有煙霧的影像,再使用本文提出的煙霧消除演算法,亦可在夜晚影像中去除部分煙霧。我們相信影像除煙會是在除霧演算法逐步成熟之後,下一個需要被進一步研究的課題。 | zh_TW |
dc.description.abstract | This thesis discusses about how hazy imaging model can be applied in many fields such as daytime dehazing, nighttime dehazing or moreover, image desmoking. The first part of the thesis is about some important existing daytime dehazing algorithm such as dark channel prior method, color line method, or negative correction model. These methods take different prior information to recover the non-hazy scene. According to these different priors, we can acquire different transmission maps and recovered results. Different priors will enhance different part of hazy images depending on the original assumption of priors and lead to different performance of different dehazing algorithm. However, these algorithms are just suitable for daytime hazy images and cannot be applied on nighttime hazy images. Nighttime hazy images usually contain artificial light source and have low luminance compared with daytime hazy images. In our second part, we introduce different nighttime dehazing algorithms including color transfer method, new imaging modeling, glowing effect removal, and image fusion based method. These algorithms pre-process nighttime hazy image to make them look like daytime hazy images and use existing daytime dehazing algorithm to recover nighttime non-hazy scenes. In our third part, we propose a novel desmoking algorithm based on hazy imaging model. We discover that there is hue distortion in smoke region of smoky images which results from unbalanced particle density distributed in each color channel. In our methods, we propose single channel dehazing based on dark channel methods to make different transmission map of each color channel. Moreover, we iteratively dehaze different color channels to make residual smoke less and less. Inspired by nighttime image dehazing, we observe nighttime smoky images share the same problem with nighttime hazy images including low lighting condition and multiple scattering process between artificial light source and particles. We propose nighttime image desmoking algorithm based on our proposed daytime desmoking algorithm. Night vision enhancement preprocessing is applied to nighttime smoky image to change it to daytime-like. Proposed daytime desmoking algorithm is then applied on enhanced images. Results shows that detail in both low luminance region and smoke region is enhanced. | en |
dc.description.provenance | Made available in DSpace on 2021-06-16T09:17:13Z (GMT). No. of bitstreams: 1 ntu-106-R04942114-1.pdf: 8653612 bytes, checksum: da595585a3a1e1f3acd1a88626367541 (MD5) Previous issue date: 2017 | en |
dc.description.tableofcontents | CONTENTS
口試委員會審定書 # 誌謝 i 中文摘要 ii ABSTRACT iii CONTENTS v LIST OF FIGURES vii Chapter 1 Introduction 1 Chapter 2 Daytime Image Dehazing Algorithm 6 2.1 Introduction 6 2.2 Dark Channel Prior Dehazing 10 2.2.1 Basic Concept of Dark Channel Prior Dehazing 10 2.2.2 Advantage and Disadvantage 13 2.3 Color Attenuation Prior Dehazing 15 2.3.1 Color Attenuation Prior 15 2.3.2 Scene Depth Restoration 17 2.3.3 Non-hazy Scene Recovery 21 2.3.4 Advantage and Disadvantage 21 2.4 Dehazing by Color Line Model 23 2.4.1 Local Color Line Model 23 2.4.2 Color Line Recovery Algorithm 24 2.4.3 Transmission Estimation by Recovered Color Line 26 2.4.4 Advantage and Disadvantage 30 Chapter 3 Nighttime Image Dehazing Algorithm 32 3.1 Introduction to Nighttime Dehazing 32 3.2 Dehazing by Color Transfer Preprocessing 35 3.2.1 Basic Concept of Dehazing by Color Transfer Preprocessing 35 3.2.2 Advantage and Disadvantage 40 3.3 New Imaging Model for Nighttime Hazy Image 42 3.3.1 Concept of New Imaging Model for Nighttime Hazy Images 42 3.3.2 Nighttime Dehazing algorithm by New Imaging Model 43 3.3.3 Advantage and Disadvantage 45 3.4 Nighttime Image Dehazing by Glowing Effect Removal 47 3.4.1 Nighttime Haze Image Model with Glowing Effect 47 3.4.2 Nighttime Dehazing Algorithm by Glowing Effect Removal 48 3.4.3 Advantage and Disadvantage 53 Chapter 4 Image Desmoking 55 4.1 Introduction 55 4.2 Limitation of Hazy Image Model 58 4.3 Atmospheric Light Detection and Estimation 65 4.4 Proposed Image Desmoking Method 69 4.5 Desmoking Result by Proposed Desmoking Method 74 4.6 Post Processing by White Balancing 77 4.7 Desmoking in Nighttime Scene 83 4.7.1 Nighttime Smoky Image Enhancement 83 4.7.2 Desmoking Results by Iterative DCP Desmoking Algorithm 89 Chapter 5 Conclusion and Future Work 92 REFERENCE 96 LIST OF FIGURES Fig. 1.1 Illustration of hazy imaging model 3 Fig. 1.2 Hazy imaging formation model description 4 Fig. 2.1 Dehazing Result using polarization (a) Minimum degree of polarization (b) Maximum degree of polarization (c) Dehazed image 7 Fig. 2.2 Dehazing result comparison between [4] [5] [6] (a) Hazy input (b) Tan’s dehazing method (c) Fattal’s dehazing method (d) He’s dehazing method 9 Fig. 2.3 Dark channel of hazy and non-hazy scene (a) non-hazy scene (b) dark channel of non-hazy scene (c) hazy scene (d) dark channel of hazy scene 11 Fig. 2.4 Position of estimated atmospheric light 12 Fig. 2.5 Estimated transmission map (a) hazy image (b) transmission map (c) refined transmission map 13 Fig. 2.6 Results by dark channel prior dehazing (a) hazy image (b) dehazed image 14 Fig. 2.7 Concentration of haze is highly correlated to saturation and intensity (a) high concentration haze region (b) medium concentration haze region (c) low concentration haze region 16 Fig. 2.8 Flow chart of dehazing method by color attenuation prior 16 Fig. 2.9 Procedure of collecting training hazy images 18 Fig. 2.10 Restored scene depth of hazy image by equation (2.7) (a) Hazy input (b) Scene depth restored by equation (2.7) (c) Scene depth with minimum filter (d) refined scene depth by guided image filter 20 Fig. 2.11 Results by color attenuation prior dehazing (a) hazy image (b) dehazed image by dark channel based method (c) dehazed image by color attenuation prior method 22 Fig. 2.12 Hazy image model combined with local color line model 24 Fig. 2.13 Procedure of color line model validation 27 Fig. 2.14 Cases of failed and successful transmission estimation by color line in which dot represents pixels in patch, purple line denotes atmospheric light and orange line denotes recovered color line (a) low intersection angle (b) large shortest distance (c) bimodel distribution (d) low shading variability (e) successful estimation in hazy patch (f) successful estimation in clear patch 28 Fig. 2.15 Transmission map estimated by color line model (a) hazy image (b) transmission estimated by local color line (c) transmission map interpolated by GMRF 29 Fig. 2.16 Dehazing results by color line estimation method (a) hazy image (b) dehazed image by dark channel based method (c) dehazed image by color line model 31 Fig. 3.1 Illumination problem in nighttime hazy images 34 Fig. 3.2 Flow chart of nighttime dehazing by color transfer preprocessing method 35 Fig. 3.3 Color transfer step between source image and target image 37 Fig. 3.4 Results of color transfer preprocessing (a) nighttime hazy image (b) target image (c) color transfer result 38 Fig. 3.5 Result of BFLCC post processing (a) dehazed image (b) luminance in (a) 40 Fig. 3.6 Other Results of nighttime dehazing (a) nighttime haze image (b) proposed by dark channel prior [3] (c) dehazed result by color transfer preprocessing 41 Fig. 3.7 Flow chart of algorithm consists of three steps including light compensation, color correction, and dehazing 43 Fig. 3.8 Results of nighttime dehazing by new imaging model (a) night hazy image (b) dehazing proposed by Pei [8] (c) dehazing by new imaging model 46 Fig. 3.9 Nighttime haze image model with glowing effect 47 Fig. 3.10 Dehazing in nighttime scene based on glowing effect removal 49 Fig. 3.11 Gradient distribution of front layer and reflection layer (a) image with reflection (b) scene without reflection (c) reflection layer (d) gradient distribution of (b) (e) gradient distribution of (c) 50 Fig. 3.12 Different λ results in different estimated L1 and L2 51 Fig. 3.13 Result of glowing removal by layer separation algorithm (a) hazy scene with glow (b) glow image (c) scene without glow 52 Fig. 3.14 Estimation of local atmospheric light (a) nighttime hazy image (b) local atmospheric light map 52 Fig. 3.15 Results of nighttime dehazing by glowing effect removal (a) night hazy image (b) dehazing proposed by Zhang [9] (c) dehazing by glowing effect removal 54 Fig. 4.1 Two main difference between hazy and smoky images including non-homogeneity and local property (a) hazy image (b) smoky image 56 Fig. 4.2 Application of desmoking algorithm (a) disaster relief (b) visual effect in movie 57 Fig. 4.3 Desmoked results directly applying dark channel prior [3] (a) smoky image (b) desmoked result 59 Fig. 4.4 Particle density distributed in RGB channel 60 Fig. 4.5 Convert from RGB color space to HSV color space 60 Fig. 4.6 HSV color model in cylindrical coordinate 62 Fig. 4.7 HSV of hazy image (a) hazy image (b) hue (c) saturation (d) value 63 Fig. 4.8 HSV of dehazed image by dark channel method from Fig4.7 (a) dehazed image (b) hue (c) saturation (d) value 63 Fig. 4.9 HSV of smoky image (a) hazy image (b) hue (c) saturation (d) value 63 Fig. 4.10 HSV of desmoked image by apply dark channel method from Fig4.9 (a) desmoked image (b) hue (c) saturation (d) value 63 Fig. 4.11 Detection of atmospheric light in hazy images 65 Fig. 4.12 Range of hue distributed by all colors 66 Fig. 4.13 Hue in smoke region (a) smoky image (b) hue of smoky image 67 Fig. 4.14 Proposed atmospheric light detection algorithm 67 Fig. 4.15 Results of thresholding on dark channel and he in smoky images (a) smoky image (b) region detected by hue (c) region detected by dark channel (d) detected smoke region 68 Fig. 4.16 Results of detected position of atmospheric light pixel 69 Fig. 4.17 One time desmoked result by single dark channel method (a) smoky image (b) red channel (c) green channel (d) blue channel 72 Fig. 4.18 Iterative dark channel prior desmoking algorithm flow chart 72 Fig. 4.19 Desmoking with proposed number of iteration (a) smoky image (b) red channel: twice (c) green channel: twice (d) blue channel: three times 73 Fig. 4.20 Desmoking result by proposed desmoking algorithm (a) smoky image (b) demsmoked image by dark channel prior based dehaze algorithm (c) desmoked image by proposed iterative single dark channel prior method 76 Fig. 4.21 Results of white balance based on gray world theory and retinex theory (a) image with warm color (b) white balance by gray world theory (c) white balance by retinex theory 78 Fig. 4.22 Results of white balance by combined gray world and retinex assumption (a) image with warm color (b) white balance by combining two assumption 79 Fig. 4.23 Desmoking result by proposed desmoking algorithm (a) smoky image (b) desmoked result without white balancing (c) desmoked results with white balancing 82 Fig. 4.24 Proposed nighttime image desmoking algorithm 83 Fig. 4.25 Example of white balancing on nighttime smoky image (a) nighttime smoky image (b) white balanced image (c)-(e) RGB response of nighttime smoky image respectively (f)-(h) RGB response of white balanced image respectively 85 Fig. 4.26 Nighttime enhancement algorithm by Dong [41] 86 Fig. 4.27 Night vision pre-processing results (a) nighttime smoky image (b) enhanced results 88 Fig. 4.28 Local atmospheric light estimation procedure 90 Fig. 4.29 Results of proposed nighttime image desmoking algorithm (a) nighttime smoky images (b) night vision enhanced images (c) desmoked results 91 | |
dc.language.iso | en | |
dc.title | 白天與夜晚影像霧霾及煙霧消除演算法之研究 | zh_TW |
dc.title | Haze Removal in Daytime and Nighttime Scene and Simple Image Desmoking by Haze Image Model | en |
dc.type | Thesis | |
dc.date.schoolyear | 105-2 | |
dc.description.degree | 碩士 | |
dc.contributor.oralexamcommittee | 鍾國亮,祁忠勇,吳家麟 | |
dc.subject.keyword | 影像除霧,夜霧影像除霧,影像除煙,影像復原, | zh_TW |
dc.subject.keyword | image dehazing,nighttime haze removal,image desmoking,image restoration, | en |
dc.relation.page | 101 | |
dc.identifier.doi | 10.6342/NTU201701442 | |
dc.rights.note | 有償授權 | |
dc.date.accepted | 2017-07-12 | |
dc.contributor.author-college | 電機資訊學院 | zh_TW |
dc.contributor.author-dept | 電信工程學研究所 | zh_TW |
顯示於系所單位: | 電信工程學研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-106-1.pdf 目前未授權公開取用 | 8.45 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。