基于局部双边滤波的实时Retinex图像增强
|
摘要
图像信息与人类社会发展息息相关,发挥着越来越重要的作用。然而,在图像信息的获取过程中,往往会受到各种不确定因素的影响,导致图像信息表达不完整,影响了图像质量。因此,通过图像增强技术处理来提高图像质量,使得图像信息完整清晰,在人类生产生活各个方面都具有十分重要的实际应用价值。
传统的图像增强算法,其基本思想是从空间域和频率域角度入手,对图像灰度图的空间域或者频率域信息进行处理。但是,在风霜雨雪等较为恶劣的自然环境下,受光照条件等因素影响,传统图像处理方法往往达不到预期的效果。基于色彩恒常性的Retinex理论,是模仿人类的视觉系统,对原始图像中的光照成分进行估计,再与原始图像进行处理,达到提高图像质量,恢复图像信息的目的。
针对传统Retinex算法的缺点,本文首先采用实时Retinex算法对图像进行分离,再利用双边滤波对图像进行局部滤波处理,在保护图像边缘不模糊受损的情况下,消除噪声,提升图片质量,也保证了图像处理的效率。最后,通过仿真实验,对传统经典Retinex图像增强方法进行分析,与其他算法的实验结果相比较,发现这种新的Retinex增强方法能够获得较好的图像处理效果,在处理时间、光晕消除和图片质量上也能够得到较好的平衡。
Image information is closely related to the development of human society, which plays a more and more important role in human's life. However, in the process of acquiring the image information, the image quality is often influenced by many uncertain factors that led to the uncompleted image information. So, the image quality is improved by the technology of image enhancement processing, and the image information becoming complete and clear, which has a very important practical application value in the human life and production.
The basic idea of traditional image enhancement algorithms is to start from the space domain and frequency domain, giving process to the image grayscale space or frequency domain information. However, in the various bad weather conditions and some particular image processing requirements, the traditional image processing methods based on human-computer interaction cannot reach the desired effect sometimes. Retinex theory, based on color constancy, is an imitation of the human visual system. The original image of light components is estimated and then processed in order to improve the image quality and recover the image information.
Aiming at the shortcoming of the traditional Retinex algorithm, this paper uses bilateral filtering to partially eliminate the noise in the image, preserving the edge without being fuzzily damaged, and then takes the real-time Retinex algorithm, not only excluding the noise, but also ensuring the efficiency of image processing at the same time. Finally, through the experiment, the traditional Retinex image enhancement methods are analyzed and compared with other filters. The experimental results prove that this new Retinex enhancement algorithm can not only improve the quality of the image, but also get a better balance in the removal of halo, the noise elimination and the short processing time.
传统的图像增强算法,其基本思想是从空间域和频率域角度入手,对图像灰度图的空间域或者频率域信息进行处理。但是,在风霜雨雪等较为恶劣的自然环境下,受光照条件等因素影响,传统图像处理方法往往达不到预期的效果。基于色彩恒常性的Retinex理论,是模仿人类的视觉系统,对原始图像中的光照成分进行估计,再与原始图像进行处理,达到提高图像质量,恢复图像信息的目的。
针对传统Retinex算法的缺点,本文首先采用实时Retinex算法对图像进行分离,再利用双边滤波对图像进行局部滤波处理,在保护图像边缘不模糊受损的情况下,消除噪声,提升图片质量,也保证了图像处理的效率。最后,通过仿真实验,对传统经典Retinex图像增强方法进行分析,与其他算法的实验结果相比较,发现这种新的Retinex增强方法能够获得较好的图像处理效果,在处理时间、光晕消除和图片质量上也能够得到较好的平衡。
Image information is closely related to the development of human society, which plays a more and more important role in human's life. However, in the process of acquiring the image information, the image quality is often influenced by many uncertain factors that led to the uncompleted image information. So, the image quality is improved by the technology of image enhancement processing, and the image information becoming complete and clear, which has a very important practical application value in the human life and production.
The basic idea of traditional image enhancement algorithms is to start from the space domain and frequency domain, giving process to the image grayscale space or frequency domain information. However, in the various bad weather conditions and some particular image processing requirements, the traditional image processing methods based on human-computer interaction cannot reach the desired effect sometimes. Retinex theory, based on color constancy, is an imitation of the human visual system. The original image of light components is estimated and then processed in order to improve the image quality and recover the image information.
Aiming at the shortcoming of the traditional Retinex algorithm, this paper uses bilateral filtering to partially eliminate the noise in the image, preserving the edge without being fuzzily damaged, and then takes the real-time Retinex algorithm, not only excluding the noise, but also ensuring the efficiency of image processing at the same time. Finally, through the experiment, the traditional Retinex image enhancement methods are analyzed and compared with other filters. The experimental results prove that this new Retinex enhancement algorithm can not only improve the quality of the image, but also get a better balance in the removal of halo, the noise elimination and the short processing time.
引文
[1]M. Eland. Retinex by two bilateral Lect Notes comput sci,2005,3459:217-229.
[2]R.C. Gonzalez, R.E. Woods.数字图像处理(第二版)[M].阮秋琦,阮宇智(译).北京：电子工业出版社,2008.
[3]章毓晋.图像工程,第二版.北京：清华大学出版社,2007.
[4]E. Land. The Retinex[J]. Amer. Scient,1964,52 (2):247-264.
[5]E. Land. Experiments in Color Vision[J]. Scientific American,1959, 200(5):84-99.
[6]T.F. Chan, Jianhong(jackie) Shen.图像处理与分析：变分,PDE,小波及随机方法[M],北京：科学出版社,2011.
[7]E. Land, J. McCann. Lightness and Retinex Theroy. J. Opt Soc. Amer.,1971, 62(1):1-11.
[8]彭玉华等.小波变换与工程应用[M].北京：科学出版社,1999.
[9]胡昌华,张伟等.基于MATLAB的系统分析与设计[M].西安：西安电子科技大学出版社,1999.
[10]张强,王正林.精通MATLAB图像处理[M].北京：电子工业出版社,2009.
[11]胡琼,汪荣贵,胡韦伟,杨万挺.基于直方图分割的彩色图像增强算法[J].中国图像图形学报,2009(9)：1776-1781.
[12]Z. Rahman. Properties of a center/surround retinex, part 1:Signal processing design. NASA Contractor Rep,1995.198194.
[13]李学明.基于Retinex理论的图像增强算法[J].计算机应用研究,2005,(2)：235-237.
[14]沈丰,赵宇明.基于实时Retinex与双边滤波的图像增强算法[J].计算机应用与软件,2009：234-238.
[15]D.J. Jobson, Z.U. Rahman, GA. Wbodell. Properties and Peformance of a center/surround retinex[J]. IEEE Transactions on Image Proeessing,1997,6(3): 451-462.
[16]B. Funt, F. Ciurea, J. McCann. Retinex in Matlab[J]. Journal of the Electronic Imaging,2004,13(1):48-57.
[17]R. Kimmel, M. Elad, D. Shaked, et al. A variational framework for retinex[J]. International Journal of Computer Vision,2003,52(1):7-23.
[18]Kang Yao, Deshou Tian. Shadow Removal from Images Using an Improved Single-Scale Retinex Color Restoration Algorithm[C]. International Joint Conference on Computational Sciences and Optimization,2009:934-938.
[19]纪则轩,陈强,孙权森等.基于双边滤波的单尺度Retinex图像增强算法[J].微电子学与计算机,2009,26(10)：99-102.
[20]D.J. Jobson. A multiscale retinex for bridging the gap between color images and the human observation of scenes[J]. IEEE Trans. Image Processing,1997,6(7): 965-976.
[21]L. Meylan, S. Susstrunk. High dynamic range image rendering with a retinex-based adaptive filter[J]. IEEE Trans. Image Processing,2006,15(9): 2820-2830.
[22]A.C. Hurlbert. Formal connection between lightness algorithms[J]. Journal of the Optical Society of America A:Optics,Image Science, and Vision,1986,3(10): 1684-1693.
[23]赵艳飞,高清维,卢一相.基于多尺度Retinex算法的遥感图像增强[J].计算机技术与发展,2008,18(2)：70-73.
[24]芮义斌,李鹏,孙锦涛.基于色彩恒常理论的图像去雾技术[J].南京理工大学学报,2006,30(5)：622-625.
[25]Z. Rahman, D.J. Jobson, GA. Woodell, et al. Automated onboard terra in analysis for precision landings[J]. Visual Information Processing XIV, Proceedings of SPIE,2006,6246:62460J.
[26]唐磊,赵春霞,王鸿南等.基于各向异性Retinex的路面图像阴影消除[J].中国图象图形学报,2008,13(2)：264-268.
[27]J. Tumblin, G. Turk. LCIS:a boundary hierarchy for detail preserving contrast reduction[C]//Computer Graphics Proceedings, Annual Conference Series, ACM SIG GRAPH, Los Angeles,1999:83-90.
[28]Wenye Mai, Jean-Michel Morel, Stanley Osher, Aichi Chien. An L1-based variational model for Retinex theory and its application to medical images[C]. IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2011: 153-160.
[29]赵志华,高清维,卢一相.一种新的基于多尺度retinex的高动态范围图像增强方法[J].电子技术,2010,11.
[30]傅剑峰,汪荣贵,张新龙,朱静.基于人眼视觉特性的Retinex算法研究[J].电子测量与仪器学报,2011,25(1)：29-37.
[31]刘家朋,赵宇明,胡福乔.基于单尺度Retinex算法的非线性图像增强算法[J].上海交通大学学报,2007,41(5)：685-688.
[32]王彦臣,李树杰,黄廉卿.基于多尺度Retinex的数字X光图像增强方法研究[J].光学精密仪器,2006,14(1)：70-76.
[33]M. Elad, R. Kimmel, D. Shaked, et al. Reduced complexity Retinex algorithm via the variational approach[J]. J Vis Commun Image R,2003,14:369-388.
[34]B. Funt, F. Ciurea. Tuning Retinex parameters[J]. Journal of Electronic Imaging, 2004,13(1):58-64.
[35]孔令同,张旭亮.GVLF及人眼灰视觉特性的研究[J].测绘通报,2001.
[36]Wu Henan, Yang Guang, Xu Zhonglin, Li Dejun, Yuan Yuwei. Remote sensing image enhancement method based on multi-scale Retinex[J].2011 International Conference of Information Technology, Computer Engineering and Management Sciences.
[37]Jae Ho Jang, Yoonsung Bae, Jong Beom Ra. Contrast-Enhanced Fusion of Multisensor Images Using Subband-Decomposed Multiscale Retinex[J]. IEEE TRANSACTIONS ON IMAGE PROCESSING,2012,21(8):3479-3490.
[38]M. Bertalmio, V. Caselles, E. Provenzi. Issues about retinex theory and contrast enhancement. Int.J.Comput.Vis., vol.83, no.1,pp.101-119,2009.
[39]S. Li, B. Yang, J. Hu. Performance comparison of different multi-resolution transforms for image fusion, Inf.Fusion, vol.12, no.2, pp.74-84, Apr.2011.
[40]Z. Rahman, D.J. Jobson, GA. Woodell. Retinex processing for automatic image enhancement. J.Electron.Imag., vol.13, no.1, pp.100-110, Jan.2004.
[41]江兴方,陶纯堪.Retinex彩色图像增强理论及其研究进展[J].光学技术,2005,31 Suppl.：258-263.
[42]肖志级,黄建军.城市彩色航空影像的边缘模糊Retinex阴影消除[J].中国体视学与图像分析,2004,9(2)：95-98.
[2]R.C. Gonzalez, R.E. Woods.数字图像处理(第二版)[M].阮秋琦,阮宇智(译).北京：电子工业出版社,2008.
[3]章毓晋.图像工程,第二版.北京：清华大学出版社,2007.
[4]E. Land. The Retinex[J]. Amer. Scient,1964,52 (2):247-264.
[5]E. Land. Experiments in Color Vision[J]. Scientific American,1959, 200(5):84-99.
[6]T.F. Chan, Jianhong(jackie) Shen.图像处理与分析：变分,PDE,小波及随机方法[M],北京：科学出版社,2011.
[7]E. Land, J. McCann. Lightness and Retinex Theroy. J. Opt Soc. Amer.,1971, 62(1):1-11.
[8]彭玉华等.小波变换与工程应用[M].北京：科学出版社,1999.
[9]胡昌华,张伟等.基于MATLAB的系统分析与设计[M].西安：西安电子科技大学出版社,1999.
[10]张强,王正林.精通MATLAB图像处理[M].北京：电子工业出版社,2009.
[11]胡琼,汪荣贵,胡韦伟,杨万挺.基于直方图分割的彩色图像增强算法[J].中国图像图形学报,2009(9)：1776-1781.
[12]Z. Rahman. Properties of a center/surround retinex, part 1:Signal processing design. NASA Contractor Rep,1995.198194.
[13]李学明.基于Retinex理论的图像增强算法[J].计算机应用研究,2005,(2)：235-237.
[14]沈丰,赵宇明.基于实时Retinex与双边滤波的图像增强算法[J].计算机应用与软件,2009：234-238.
[15]D.J. Jobson, Z.U. Rahman, GA. Wbodell. Properties and Peformance of a center/surround retinex[J]. IEEE Transactions on Image Proeessing,1997,6(3): 451-462.
[16]B. Funt, F. Ciurea, J. McCann. Retinex in Matlab[J]. Journal of the Electronic Imaging,2004,13(1):48-57.
[17]R. Kimmel, M. Elad, D. Shaked, et al. A variational framework for retinex[J]. International Journal of Computer Vision,2003,52(1):7-23.
[18]Kang Yao, Deshou Tian. Shadow Removal from Images Using an Improved Single-Scale Retinex Color Restoration Algorithm[C]. International Joint Conference on Computational Sciences and Optimization,2009:934-938.
[19]纪则轩,陈强,孙权森等.基于双边滤波的单尺度Retinex图像增强算法[J].微电子学与计算机,2009,26(10)：99-102.
[20]D.J. Jobson. A multiscale retinex for bridging the gap between color images and the human observation of scenes[J]. IEEE Trans. Image Processing,1997,6(7): 965-976.
[21]L. Meylan, S. Susstrunk. High dynamic range image rendering with a retinex-based adaptive filter[J]. IEEE Trans. Image Processing,2006,15(9): 2820-2830.
[22]A.C. Hurlbert. Formal connection between lightness algorithms[J]. Journal of the Optical Society of America A:Optics,Image Science, and Vision,1986,3(10): 1684-1693.
[23]赵艳飞,高清维,卢一相.基于多尺度Retinex算法的遥感图像增强[J].计算机技术与发展,2008,18(2)：70-73.
[24]芮义斌,李鹏,孙锦涛.基于色彩恒常理论的图像去雾技术[J].南京理工大学学报,2006,30(5)：622-625.
[25]Z. Rahman, D.J. Jobson, GA. Woodell, et al. Automated onboard terra in analysis for precision landings[J]. Visual Information Processing XIV, Proceedings of SPIE,2006,6246:62460J.
[26]唐磊,赵春霞,王鸿南等.基于各向异性Retinex的路面图像阴影消除[J].中国图象图形学报,2008,13(2)：264-268.
[27]J. Tumblin, G. Turk. LCIS:a boundary hierarchy for detail preserving contrast reduction[C]//Computer Graphics Proceedings, Annual Conference Series, ACM SIG GRAPH, Los Angeles,1999:83-90.
[28]Wenye Mai, Jean-Michel Morel, Stanley Osher, Aichi Chien. An L1-based variational model for Retinex theory and its application to medical images[C]. IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2011: 153-160.
[29]赵志华,高清维,卢一相.一种新的基于多尺度retinex的高动态范围图像增强方法[J].电子技术,2010,11.
[30]傅剑峰,汪荣贵,张新龙,朱静.基于人眼视觉特性的Retinex算法研究[J].电子测量与仪器学报,2011,25(1)：29-37.
[31]刘家朋,赵宇明,胡福乔.基于单尺度Retinex算法的非线性图像增强算法[J].上海交通大学学报,2007,41(5)：685-688.
[32]王彦臣,李树杰,黄廉卿.基于多尺度Retinex的数字X光图像增强方法研究[J].光学精密仪器,2006,14(1)：70-76.
[33]M. Elad, R. Kimmel, D. Shaked, et al. Reduced complexity Retinex algorithm via the variational approach[J]. J Vis Commun Image R,2003,14:369-388.
[34]B. Funt, F. Ciurea. Tuning Retinex parameters[J]. Journal of Electronic Imaging, 2004,13(1):58-64.
[35]孔令同,张旭亮.GVLF及人眼灰视觉特性的研究[J].测绘通报,2001.
[36]Wu Henan, Yang Guang, Xu Zhonglin, Li Dejun, Yuan Yuwei. Remote sensing image enhancement method based on multi-scale Retinex[J].2011 International Conference of Information Technology, Computer Engineering and Management Sciences.
[37]Jae Ho Jang, Yoonsung Bae, Jong Beom Ra. Contrast-Enhanced Fusion of Multisensor Images Using Subband-Decomposed Multiscale Retinex[J]. IEEE TRANSACTIONS ON IMAGE PROCESSING,2012,21(8):3479-3490.
[38]M. Bertalmio, V. Caselles, E. Provenzi. Issues about retinex theory and contrast enhancement. Int.J.Comput.Vis., vol.83, no.1,pp.101-119,2009.
[39]S. Li, B. Yang, J. Hu. Performance comparison of different multi-resolution transforms for image fusion, Inf.Fusion, vol.12, no.2, pp.74-84, Apr.2011.
[40]Z. Rahman, D.J. Jobson, GA. Woodell. Retinex processing for automatic image enhancement. J.Electron.Imag., vol.13, no.1, pp.100-110, Jan.2004.
[41]江兴方,陶纯堪.Retinex彩色图像增强理论及其研究进展[J].光学技术,2005,31 Suppl.：258-263.
[42]肖志级,黄建军.城市彩色航空影像的边缘模糊Retinex阴影消除[J].中国体视学与图像分析,2004,9(2)：95-98.