摘要
为了便于人眼视觉观察和后续计算机分析处理,图像复原和图像增强能够突出图像中的部分细节信息。由于存在大气粒子的作用,雨雾天气时,空气中充满着小水珠,各种图像实物的反射光与周围环境光被散射或者吸收,使得到达接收装置的图像信息变得模糊不清,同时现有视频监控、目标跟踪器件等对天气十分敏感,大气中这些粒子严重影响了光学多媒体设备的应用。
雾天图像的增强复原问题越来越受到科学家们的关注,对其的研究也越演越热。日常生活中,希望对雾引起的不良效果进行改善的需求越来越大。由于雾天物体的成像过程十分复杂,想要给出统一的描述十分困难,因此,雾天图像增强处理问题亟待解决。本文主要研究以下内容：
(1)研究出基于MSR的雾天数字图像增强的处理算法。由于传统卷积函数决定了Retinex算法最终处理效果,且处理结果效果很差。究其缘由是早期算法的卷积滤波函数是同态Gauss滤波器。本文采用自适应滤波器,即当图像灰度在某阈值内时使用同态滤波器处理图像；超过这一阈值,则采用各项异性滤波器。同时,传统多尺度Retinex使用相同的权重因子,本文在自适应滤波器的方法上结合不同的权重因子,实验证明图像细节特征得到增强,对比度明显提高,改善了传统算法的失真问题和光晕现象。
(2)采用基于Retinex算法增强图像的修正的方法,运用非线性方法修正图像的反射分量和照射分量。采用全局对比度增强函数对图像的照射分量进行强度拉伸,提高整体对比效果。反射分量使用非线性S型函数进行变换,因其能较大的改变中间值,而对偏大和偏小的值改变不大,起到增强图像局部对比度的目的。与传统MSR(多尺度Retinex)算法比较,结果显示图像对比度明显增强,视觉效果得到改善,提升了图像中明暗凸变部分。且算法处理速度较快,处理结果不会出现明显的泛白或失真现象。
Image restoration and enhancement can highlight some details of images, making it easier for observation and post-computer analysis and processing. Due to atmospheric particles, the air is filled with droplets in rainy and foggy weather, thereby the reflected light of real objects and ambient light are either scattered or absorbed, blurring images received by the receiver; at the same time, being sensitive to the weather, such optical media devices as video surveillances and target trackers are sensitively affected by the existence of those particles in the air.
The enhancement and restoration of images in foggy weather is attracting more and more attention from scientists, and the relevant studies are now on the rise. It is earnestly demanded that adverse results caused by fog be the demand of the adverse effects caused by fog be removed. The sophisticated imaging process of objects in fog itself makes it difficult to produce a comprehensive description, so it still remains as a problem to be solved to process images in foggy weather.
The main content of the paper contains the following:
(1) An improved Retinex rainy day digital image enhancement method is proposed. Because the traditional convolution function determines the results of Retinex algorithm and the results poorly. Investigating its reason, we find it is the convolution algorithm is homomorphic filter function Gauss filter. In this paper, a self-adaption filter is adopted, when the image gray levels within a certain threshold we use a homomorphic filter processing the image;else it exceeds this threshold value, we use anisotropic filter. Meanwhile, the traditional multi-scale Retinex using the same weighting factor, my paper uses the self-adaption filter method with different weighting factors. Experiments show that this method enhances image detail features,;the contrast of image is markedly improved,; and the distortion and hole phenomena have been improved.
(2) Study on Retinex image enhancement algorithm based on the modified method. Using non-linear methods modify the reflection and irradiation component of an image. Using a global contrast function to stretching the illumination of the image, it improves the overall contrast. A S-type function is used to alter the reflection of an image, it can largely changes the Intermediate value, while the higher and lower values change little, play a local image contrast enhancement purposes. Compared with traditional MSR (multi-scale Retinex) algorithm, the results show significantly enhancement, visual effects are improved, and the bright and dim part of the image is enhanced. Meanwhile, the algorithm processing fleetly, and the results is not present apparently whitening or distortion.
引文
[1]Oakley J P, Satherley B L. Improving image quality in poor visibility conditions using a physical model for contrast degradation[J]. Image Processing, IEEE Transactions on,1998,7(2):167-179.
[2]Nayar S K, Narasimhan S G. Vision in bad weather[C]//Computer Vision,1999. The Proceedings of the Seventh IEEE International Conference on. IEEE,1999,2: 820-827.
[3]Land E H, McCann J J. Lightness and retinex theory[J]. Journal of the Optical society of America,1971,61(1):1-11.
[4]Land E H. The retinex theory of color vision[M]. Scientific America.,1977.
[5]Orsini G, Ramponi G, Carrai P, et al. A modified Retinex for image contrast enhancement and dynamics control[C]//Image Processing,2003. ICIP 2003. Proceedings.2003 International Conference on. IEEE,2003,3:Ⅲ-393-6 vol.2.
[6]Jensen J R. Introductory digital image processing:a remote sensing perspective[R]. Univ. of South Carolina, Columbus,1986.
[7]Jang J H, Choi B, Kim S D, et al. Sub-band decomposed multiscale retinex with space varying gain[C]//Image Processing,2008. ICIP 2008.15th IEEE International Conference on. IEEE,2008:3168-3171.
[8]Wang Q, Ward R K. Fast image/video contrast enhancement based on weighted thresholded histogram equalization[J]. Consumer Electronics, IEEE Transactions on,2007,53(2):757-764.
[9]刘锐.基于暗原色先验的一种快速图像去雾方法[J].计算机光盘软件与应用,2012(21).
[10]Rahman Z, Jobson D J, Woodell G A. Multi-scale retinex for color image enhancement[C]//Image Processing,1996. Proceedings., International Conference on. IEEE,1996,3:1003-1006.
[11]Levin A, Lischinski D, Weiss Y. A closed-form solution to natural image matting[J]. Pattern Analysis and Machine Intelligence, IEEE Transactions on, 2008,30(2):228-242.
[12]Hautiere N, Tarel J P, Aubert D, et al. Blind contrast enhancement assessment by gradient ratioing at visible edges[J]. Image Analysis & Stereology Journal,2008, 27(2):87-95.
[13]Kocak D M, Dalgleish F R, Caimi F M, et al. A focus on recent developments and trends in underwater imaging[J]. Marine Technology Society Journal,2008,42(1): 52.
[14]Narasimhan S G, Nayar S K. Contrast restoration of weather degraded images[J]. Pattern Analysis and Machine Intelligence, IEEE Transactions on,2003,25(6): 713-724.
[15]Von Kries R, Koletzko B, Sauerwald T, et al. Breast feeding and obesity:cross sectional study [J]. BMJ:British Medical Journal,1999,319(7203):147.
[16]Brelstaff G, Blake A. Computing lightness[J]. Pattern Recognition Letters,1987, 5(2):129-138.
[17]Rahman Z, Jobson D J, Woodell G A. Multi-scale retinex for color image enhancement[C]//Image Processing,1996. Proceedings., International Conference on. IEEE,1996,3:1003-1006.
[18]Rahman Z, Jobson D J, Woodell G A. Retinex processing for automatic image enhancement[J]. Journal of Electronic Imaging,2004,13(1):100-110.
[19]Frankle J A, McCann J J. Method and apparatus for lightness imaging:U.S. Patent 4,384,336[P].1983-5-17.
[20]Funt B, Ciurea F, McCann J. Retinex in matlab[C]//Proceedings of the IS&T/SID Eighth Color Imaging Conference:Color Science, Systems and Applications.2000: 112-121.
[21]Ciurea F, Funt B. Tuning retinex parameters [J]. Journal of Electronic Imaging, 2004,13(1):58-64.
[22]Tumblin J, Turk G. LCIS:A boundary hierarchy for detail-preserving contrast reduction[C]//Proceedings of the 26th annual conference on Computer graphics and interactive techniques. ACM Press/Addison-Wesley Publishing Co.,1999: 83-90.
[23]Jang J H, Choi B, Kim S D, et al. Sub-band decomposed multiscale retinex with space varying gain[C]//Image Processing,2008. ICIP 2008.15th IEEE International Conference on. IEEE,2008:3168-3171.
[24]汪荣贵,张璇,张新龙,等.一种新型自适应Retinex图像增强方法研究[J].电子学报,2010,38(12)：2933-2936.
[25]Moore A, Allman J, Goodman R M. A real-time neural system for color constancy[J]. Neural Networks,IEEE Transactions on,1991,2(2):237-247.
[26]Hurlbert A. Formal connections between lightness algorithms[J]. JOSA A,1986, 3(10):1684-1693.
[27]Mamdani E H, Assilian S. An experiment in linguistic synthesis with a fuzzy logic controller[J]. International journal of man-machine studies,1975,7(1):1-13.
[28]Land E H. Recent advances in retinex theory and some implications for cortical computations:color vision and the natural image[J]. Proceedings of the National Academy of Sciences of the United States of America,1983,80(16):5163.
[29]Zhai Y S, Liu X M. An improved fog-degraded image enhancement algorithm[C]//Wavelet Analysis and Pattern Recognition,2007. ICWAPR'07. International Conference on. IEEE,2007,2:522-526.
[30]李冠章,罗武胜,李沛.一种高效地修正Retinex图像自适应对比度增强算法[J].测试技术学报,2009,23(5)：445-451.
[31]Hautiere N, Tarel J P, Aubert D, et al. Blind contrast enhancement assessment by gradient ratioing at visible edges[J]. Image Analysis & Stereology Journal,2008, 27(2):87-95.
[32]Kim K, Davis L. A fine-structure image/video quality measure using local statistics[C]//Image Processing,2004. ICIP'04.2004 International Conference on. IEEE,2004,5:3535-3538.
[33]Abdou IE, Dusaussoy N J. Survey of image quality measurements[C] //Proceedings of 1986 ACM Fall joint computer conference. IEEE Computer Society Press,1986:71-78.
[34]Bjontegard G. Calculation of average PSNR differences between RD-curves[J]. ITU-T VCEG-M33,2001.
[35]Silver W M, Wolff R A, Dynneson R E. Digital image processing system:U.S. Patent 4,972,359[P].1990-11-20.
[36]McCartney E J. Optics of the Atmosphere:Scattering by molecules and particles[J]. New York, John Wiley and Sons, Inc.,1976.421 p.,1976,1.
[37]Chen M, Men A, Fan P, et al. Single image defogging[C]//Network Infrastructure and Digital Content,2009. IC-NIDC 2009. IEEE International Conference on. IEEE,2009:675-679.
[38]Xu H, Guo J, Liu Q, et al. Fast image dehazing using improved dark channel prior[C]//Information Science and Technology (ICIST),2012 International Conference on. IEEE,2012:663-667.
[39]Kim K, Davis L. A fine-structure image/video quality measure using local statistics[C]//Image Processing,2004. ICIP'04.2004 International Conference on. IEEE,2004,5:3535-3538.
[40]嵇晓强,戴明,孙丽娜,等.暗原色先验图像去雾算法研究[J].光电子.激光,2011,22(6)：926-930.
[41]Greene B, Levin J. Dark energy and stabilization of extra dimensions[J]. Journal of High Energy Physics,2007,2007(11):096.
[42]Tan R T. Visibility in bad weather from a single image[C]//Computer Vision and Pattern Recognition,2008. CVPR 2008. IEEE Conference on. IEEE,2008:1-8.
[43]胡学友.雾天降质图像的增强复原算法研究[D].安徽大学,2011.
[44]Louwe R J W, McDermott L N, Sonke J J, et al. The long-term stability of amorphous silicon flat panel imaging devices for dosimetry purposes[J]. Medical physics,2004,31:2989.
[45]方帅,王勇,曹洋,等.单幅雾天图像复原[J].电子学报,2010,38(10)：2279.