臺灣博碩士論文加值系統

隨著民眾安全意識抬頭，各種監控技術興起，其中尤以監視器及行車紀錄器使用最為頻繁，無非是想透過影像忠實地紀錄下人們行為舉止，不過，監視系統十分仰賴充足光源，因此監視器於低光照環境下成效不彰，而犯罪及事故普遍發生於視線不佳之夜間，有鑑於此，提升夜間影像品質之演算法至關重大，當事故發生時，才能透過影像還原真相，還被害者公道。
悉知影像處理技術皆建立於光線充足之前提下，但就實際應用而言，影像處理領域反而欠缺還原夜間影像之強化技術。以Dark Channel Prior (DCP) 除霧演算法為例，其方法雖然對於日間影像除霧有卓越成效，可惜對於夜間影像並不適用，主因為日間景物受到日光均勻照射，故能以物體受光量結合大氣散射物理模型反推景深；但夜間影像並無日光，反而是以劇烈且不穩定之人造光源 (如路燈、車燈、交通號誌) 照明，打破了DCP之假設前提，故不能直接使用。本文結合消除光暈、DCP及色彩校正等影像處理技術，提出一種有效強化低光照環境影像之方法。
本文基於Dark Channel Prior演算法，利用負片系統，將DCP消除濃霧之特性，轉化成對暗處影像之光源補償，藉此強化低光照影像。首先，需要移除人造光源，修正受人造光源影響之區域，再將影像轉成負片，利用DCP推算物體景深 (Scene Depth) 及透射率 (Transmission)，並使用導向濾波器 (Guided Image Filtering) 使透射率更趨細緻，再透過大氣散射物理模型補償光源，最後，將影像由負片轉正，對補償後影像進行色彩校正，即可有效強化低光照環境之影像。

Nowadays as people have higher awareness of public safety, different kinds of surveillance technologies emerge. Amongst these surveillance technologies, closed-circuit television (CCTV) and car video recorder are most frequently used by people, who simply want to honestly record the behaviors and acts of other people through images. However, surveillance system highly relies on sufficient light, so surveillance equipment seems ineffective in a low-light environment. But crimes and incidents commonly happen in the nighttime with poor visibility. Therefore, the algorithm for enhancement of nighttime image quality is extremely important. With a sound algorithm, unclear images obtained in an incident can be restored by image enhancement, and the truth can thus be known from clearer images, letting the victims receive fair treatment.
It is known that image processing technique is established on a prerequisite of sufficient light. However, in times of practical application, the area of image processing is in lack of image enhancement technique to restore nighttime image. Taking the dehazing algorithm, dark channel prior (DCP) for example, although this method has remarkable effect in dehazing of daytime image, it is completely inapplicable to nighttime image. This is mainly because daytime scene or object has sunlight evenly shining on, so that the light amount received by the scene or object can combine with the physical model of atmospheric scattering to inversely derive the scene depth. Nevertheless, nighttime image has no sunlight to shine on, but is irradiated by intense and unstable artificial light (such as streetlamp, car light, and traffic lights), breaking the supposed prerequisite of DCP and disabling its direct application to nighttime image. The paper combines the image processing techniques of glow removal, DCP and color calibration in order to develop an effective method to enhance images in low-light environment.
Based on DCP algorithm , the paper uses negative film system and DCP’s dehazing feature to perform light compensation for the dark area of image, and employs this way to enhance low-light image. First of all, artificial light is removed; the area affected by artificial light is corrected; and the image is turned to be negative film. Using DCP, the paper calculates the scene depth and transmission of object, and uses guided image filtering to make transmission more refined. Through the physical model of atmospheric scattering, light is compensated for the image. Finally, the negative film of image is converted to be positive, and color calibration of image is carried out after compensation. Then the image in low-light environment is effectively enhanced.

論文審定書 i
論文公開授權書 ii
中文摘要 iii
英文摘要 iv
目 錄 vi
圖　次 vii
第一章 緒論 1
1.1 大氣物理散射模型 3
1.2 Dark Channel Prior 5
第二章 相關研究 9
2.1 低光照影像與含霧影像關聯性 9
2.2 Dark Channel Prior除霧演算法於夜間影像之可行性 10
2.3 模擬光暈 13
2.4 增強夜間影像 15
第三章 研究方法 16
3.1 移除光暈 18
3.2 低光照影像增強 21
3.3 色彩校正 26
第四章 實驗結果 28
第五章 結論 39
參考文獻 41

[1]Y. Li, Robby T. Tan, Michael S. Brown, “Nighttime haze removal with glow and multiple light colors,” Proc. 2014 IEEE Int''l Conf. on Comput. Vis., pp. 226-234, 2015.
[2]J. T. Houghton, “The physics of atmospheres,” Chapter 2, 2nd ed. Cambridge University Press, 2001.
[3]K. He, J. Sun, and X. Tang, “Single image haze removal using dark channel prior,” IEEE Trans. on Pattern Anal. and Mach. Intell., vol. 33, no. 12, pp. 2341-2353, 2011.
[4]K. He, J. Sun, and X. Tang, “Guided image filtering,” IEEE Trans. on Pattern Anal. and Mach. Intell., vol. 35, no. 6, pp. 1397-1409, 2013.
[5]J. Yu, D.P. Li, “Physics-Based fast single image fog removal,” Acta Automatica Sinica, 2011.
[6]R. Fattal, “Single image dehazing,” ACM Trans. on Graphics, vol. 27, no. 3, pp. 1-9, 2008.
[7]R. Tan, “Visibility in bad weather from a single image,” Proc.2008 IEEE Int''l Conf. on CVPR, pp. 1-8, 2008.
[8]S. G. Narasimhan, and S. K. Nayar, “Contrast restoration of weather degraded images,” IEEE Trans. on Pattern Anal. and Mach. Intell., vol. 25, no. 6, pp. 713-724, 2003.
[9]S. K. Nayar, and S. G. Narasimhan, “Vision in bad weather,” Proc. Seventh IEEE Int’l Conf. on Comput. Vis., vol. 2, pp. 820–827, 1999.
[10]X. J. Huang, Y. D. Lai, F. Chien, “Fast single image haze removal algorithm,” Journal of Computer applications, pp. 3028-3030, 2010.
[11]Y. Y. Schechner, S. G. Narasimhan, and S. K. Nayar, “Instant dehazing of images using polarization,” Proc. 2001 IEEE Int''l Conf. on CVPR, vol. 1, pp. 325–332, 2001.
[12]K. Lebart, C. Smith, E. Trucco, and D. M. Lane, “Automatic indexing of underwater survey video: algorithm and benchmarking method,” IEEE J. Ocean. Eng., Vol. 28, No. 4, pp. 673 - 686, 2003.
[13]L. Chao, and M. Wang, “Removal of water scattering,” Int’l Conf. on Computer Engineering and Technology, Vol. 2, pp. 35-39, 2010.
[14]N. Carlevaris-Bianco, A. Mohan, and R. M. Eustice, “Initial results in underwater single image dehazing,” Proc. IEEE OCEANS, pp. 1-8, 2010.
[15]J.Y. Chiang, and Y.C. Chen, “Underwater image enhancement by wavelength compensation and dehazing,” IEEE Trans. on Image Proc., vol. 21, no. 4, pp. 1756-1769, 2012.
[16]X. Dong, G. Wang, Y. A. Pang, W. Li, J. G. Wen, W. Meng, and Y. Lu, “Fast efficient algorithm for enhancement of low lighting video,” Proc. 2011 IEEE Int''l Conf. on Multimedia and Expo, pp. 1-6, 2011.
[17]S. G. Narasimhan, and S. K. Nayar. “Shedding light on the weather,” Proc. 2003 IEEE Int''l Conf. on CVPR, 2003.
[18]N. M. Kwok, D. Wang, X. Jia, S. Y. Chen, G. Fang, and Q. P. Ha, “Gray world based color correction and intensity preservation for image enhancement,” Proc. 2011 IEEE Int''l Conf. on CISP, Vol. 2, pp. 994-998, 2011.
[19]G. Meng, Y. Wang, J. Duan, S. Xiang, and C. Pan. “Efficient image dehazing with boundary constraint and contextual regularization,” Proc. 2013 IEEE Int''l Conf. on Comput. Vis., pp. 617-624, 2013.
[20]J. Zhang, Y. Cao, and Z. Wang. “Nighttime haze removal based on a new imaging model,” Proc. 2014 IEEE Int''l Conf. on Image Proc., 2014.