|
|
|
发布时间: 2017-10-16 |
图像处理和编码 |
|
|
|
毛东月, 谢正祥, 贺向前, 贾媛媛, 周丽华 |
收稿日期: 2017-03-20; 修回日期: 2017-06-20
基金项目: 重庆市科委基金项目（cstc2013yykfA10013）
第一作者简介:
毛东月(1989-), 男, 重庆医科大学生物医学工程专业硕士研究生, 主要研究方向为医学图像处理。E-mail:dongyue4544@163.com.
中图法分类号: TP391
文献标识码: A
文章编号: 1006-8961(2017)10-1356-08
|
摘要
目的 在低照度环境下，由于受图像采集设备的限制，导致获取到的图像往往亮度低、对比度差。针对这一问题，提出一种自适应双向保带宽对数变换的增强算法。方法 首先通过标准化变换将低照度图像处理成标准化图像，然后根据标准化图像的平均亮度进行自适应双向保带宽对数变换，最后对图像取整输出，从而得到增强后的图像。结果 实验选用LIVE database release2标准库中29幅高质量图像作为参考图像，然后经Photoshop CS5统一处理成低照度图像，使用本文算法对其增强，并与直方图均衡化（HE）、多尺度Retinex增强（MSR）、自然保持的增强算法（NPEA）的结果进行比较。本文算法增强后的图像其整体对比度和亮度在主观上都有较大提高，增强效果优于其他3种方法；同时，本文算法峰值信噪比（PSNR）和结构相似度（SSIM）客观指标平均值分别为22.75和0.86，明显高于其他3种算法。另外，在算法运行效率方面，本文算法平均运行时间也较短，约为74 ms。结论 本文算法增强后的图像更自然、更符合人眼视觉特性，且算法简单易于实现，运行效率高。该算法广泛适用于背光或光照不均的低照度环境下的图像增强。
关键词
低照度; 图像增强; 标准化变换; 自适应双向对数变换; 保带宽
Mao Dongyue, Xie Zhengxiang, He Xiangqian, Jia Yuanyuan, Zhou Lihua Abstract
Objective In a low-illumination environment, such as in nighttime video surveillance and some special scenes, the limitations of the image acquisition device, non-professional photography, and loss of information in video transmission often results in the acquisition of image with low brightness and poor contrast.Such conditions bring great challenges to image post-processing, such as image recognition, segmentation, and classification.Therefore, enhancing a low-illumination image in the preprocessing step is necessary.The aim of low-illumination image enhancement is to enhance the dark area, suppress the highlighted area, and realize the image clarity.At present, most of the methods are based on Histogram Equalization(HE), Retinex theory, and homomorphic filtering.Particularly, HE can adaptively improve the dynamic range of the image gray scale but it also can lead to unnatural over-enhancement of image contrast; and the merging gray levels cause loss of some details of the image.Retinex-based algorithms can enhance image contrast to a certain extent, but the computational complexity is generally high, the computational speed is slow, and the image color is also easily distorted.The premise of the enhancement algorithm based on homomorphic filter is that the illumination is uniform, so this method is unsuitable for low-illumination images with uneven illumination.Moreover, this method also lacks self-adaptability because the dynamic range depends on the frequency of the filter.Although the logarithmic transform can show more details of the dark area, but it also loses some details of the bright region.The Retinex algorithm and the enhancement algorithm based on homomorphic filter all involve logarithmic transformation, but none of the logarithmic base are specified.Only one-way logarithmic transformation is carried out, which can only improve the image contrast of the dark area.To overcome the shortcomings of the existing algorithms, and inspired by the characteristics of logarithmic transformation, this paper proposes a low-illumination image enhancement algorithm based on adaptive bilateral logarithm transformation with bandwidth preserving. Method The proposed method includes four steps.First, the low-illumination image is transformed into a standardized image by a special gray transformation called the standard transformation, which can stretch the image contrast to some extent.The purpose of image standard transformation is to make the image gray/color spectrum width equal to 256 to preserve full bandwidth.Compared with non-standardized image, the contrast and brightness of standardized image had increased.Standard transformation lays foundation for further image quality optimization.The second step of the algorithm is computation of the Average Luminance(AL) of the standardized image.Then, the adaptive bilateral logarithm transformation with preserved bandwidth is performed according to AL.More concretely, if AL is less than 127.5, reverse logarithm transformation with bandwidth preserving is carried out first, and then the forward logarithm transformation with bandwidth preserving is performed.Otherwise, forward logarithm transformation with bandwidth preserving is carried out, and then the reverse logarithm transformation with bandwidth preserving is performed.Through calculation, the value of logarithmic base is set to 1.021 983 956 89, thus achieving logarithm transformation with bandwidth preserving.Through this step, image details both in dark area and bright area can be displayed.Finally, the image is rounded out to obtain the enhanced image. Result In the experiment, 29 high-quality images in the LIVE database release 2 are used as reference images, and then processed into low-illumination images by Photoshop CS5.After that, the proposed algorithm is utilized to enhance these low-illumination images and compared with the enhanced results obtained by HE, Multi-scale Retinex(MSR), and Natural Preserved Enhancement Algorithm(NPEA).Qualitative and quantitative analyses are conducted to evaluate the proposed algorithm.Experimental results show that the overall contrast and brightness of the proposed method are improved subjectively, and the enhancement effect is better compared with the other three enhancement algorithms.Simultaneously, the Peak Signal-to-Noise Ratio(PSNR) and Structure Similarity(SSIM) value obtained by the proposed method is higher than the other three algorithms.The average PSNR and SSIM values obtained by the proposed method is 22.75 dB and 0.86, whereas the average PSNR and SSIM value of the other three method are 16.16 dB and 0.58(HE), 15.82 dB, and 0.62(MSR), 18.62 dB, and 0.78(NPEA), respectively.In addition, the average running time of the proposed algorithm is relatively short(~74 ms); however, the running time of MSR and NPEA are respectively 11.28 s and 11.58 s under the same conditions. Conclusion The proposed method makes up for the defects of retinex algorithm and homomorphic filtering method, which can improve the dark area and bright area contrast of the image at the same time.Consequently, it can enhance the low-illumination image effectively.Moreover, the algorithm can eliminate the halo artifact caused by Retinex, and it does not merge gray levels as HE.The enhanced image is more natural and more consistent with the human visual system.Meanwhile, the proposed algorithm is simple and easy to implement, which can greatly improve the operational efficiency.The proposed method can be widely applied to image enhancement in low-illumination environment under backlight or uneven illumination.However, the limitation of the proposed algorithm is the contrast and brightness of the enhanced image should be further improved.The future work will focus on applying the average luminance transformation to the enhanced image to do further enhancement.In addition, for the sake of further improving the robustness of the algorithm, more tests and verification are required for the nighttime video monitoring field.
Key words
low illumination; image enhancement; standard transformation; adaptive bilateral logarithm transformation; bandwidth preserving
0 引言
目前数字图像处理技术已广泛用于生物医学成像、安防监控、计算机视觉、遥感成像、天气预报、雷达监测、军事侦察等领域。但在低照度环境下，受图像采集设备的限制，导致获取到的图像往往偏暗、对比度差，这给图像识别、分割、分类等后处理工作带来了极大挑战。因此，对低照度图像做增强预处理很有必要。
现有低照度图像增强算法主要有基于像素的灰度变换法[1-3]、基于人类颜色恒常性的Retinex增强算法[4-8]、基于同态滤波的增强算法[9-10]等。基于像素的灰度变换法常见的有直方图均衡化(HE)、对数变换、伽马变换等，其中直方图均衡化容易导致图像对比度不自然的过增强，且变换后的图像灰度级合并会导致图像某些细节丢失[1]；对数变换可扩展图像暗区像素值但压缩图像亮区像素值[11]，因此对数变换虽能展现更多图像暗区细节但同时也会丢失部分亮区细节信息。Retinex增强算法其基本思想是在对数域，将乘性的照射分量与反射分量变为加性，然后设法估计照射分量并去除，得到反映图像本质的反射分量，最后进行对比度拉伸输出。主要的Retinex算法有单尺度Retinex算法(SSR)[4]、多尺度Retinex算法(MSR)[5]以及在SSR、MSR基础上的改进算法[6-8]，这些算法在一定程度上可增强图像对比度，但计算复杂度普遍较高，运行速度慢，图像颜色也易失真。基于同态滤波的增强算法其前提条件是假设光照均匀，因此对于光照不均的低照度图像，增强效果不好。Retinex算法和基于同态滤波的增强算法都涉及对数变换，但都没有规定对数变换的底数，都仅是做了单向的对数变换，因而也只能提高图像暗区对比度。
针对现有算法存在的缺陷，本文受到图像标准化变换[12]及对数变换特点的启发，提出一种基于自适应双向保带宽对数变换的低照度图像增强算法。该算法首先对低照度图像做标准化变换，再根据标准化图像的平均亮度(AL)进行自适应双向保带宽对数变换，最后对图像取整输出得到增强后图像。实验结果表明，本文算法增强后的图像更自然，无光晕伪影，更符合人眼视觉特性，且该算法计算复杂度较低，运行效率更高。
1 理论基础
1.1 灰度/色度谱分级平坦化理论
灰度/色度谱分级平坦化理论是对传统灰度直方图的改进，主要用来挖掘图像中少数像素点具有的灰度信息，在灰度/色度谱的分析中起到很大作用。灰度/色度谱分级平坦化理论公式[13]为
| $ T\left( g \right)=\frac{{{O}^{\frac{1}{m}}}\left( g \right)\sum\limits_{g=0}^{255}{O\left( g \right)}}{\sum\limits_{g=0}^{255}{{{O}^{\frac{1}{m}}}\left( g \right)}} $ | (1) |
式中，
灰度/色度谱分级平坦化对图片的处理效果如图 1所示。其中图 1(a)为原图(注：原图来自matlab2016a安装目录下的pout.tif)，由于分布在高端灰度级(即图像亮区)的像素个数较少，使得这部分像素所在的灰度级在传统归一化灰度直方图(也即图 1(b)1级平坦化灰度谱)上不易被观察到，因此无法了解整幅图的灰度分布情况。而图 1(a)pout经过4级平坦化后，如图 1(c)所示，200灰度值附近是有灰度信息分布的。但不管平坦化级数有多高，灰度/色度谱分级平坦化都不会增加新的灰度信息。
1.2 图像标准化变换
图像标准化变换(standard transformation)[12]本质是一种特殊的灰度变换，它是Zadeh-X变换[14]的一种特殊情况，其公式定义为
| $ {{T}_{\text{std}}}\left( i, x, y \right)=255\times \frac{O\left( i, x, y \right)-\max \left[\min \left( \text{R} \right), \min \left( \text{G} \right), \min \left( \text{B} \right) \right]}{\min \left[\max \left( \text{R} \right), \max \left( \text{G} \right), \max \left( \text{B} \right) \right]-\max \left[\min \left[\text{R} \right], \min \left( \text{G} \right), \min \left( \text{B} \right) \right]} $ | (2) |
式中，
图像标准化变换的目的是使图像的灰度/色度谱的宽度达到256灰度级；相比未标准化的图像，标准化后图像的对比度和亮度增大，从而使图像具有更好的视觉效果[15]。标准化图像为双向保带宽对数变换等图像质量优化奠定了基础。如图 2所示，图 2(a)为低亮度图像，其R、G、B分量直方图 4级色度谱(即图 2(c))较窄，经过标准化变换后图像色度谱(图 2(d))带宽扩展到0 255，图像亮度、对比度也得到明显改善(图 2(b))。
1.3 保带宽对数变换
对于标准化图像，灰度/色度谱带宽为256，其低端灰度/色度值等于0，高端灰度/色度值等于255。为使对数变换有意义，需要对标准化图像各像素值统一加1，这样图像灰度/色度值的取值范围从0 255变为1 256。这里定义变换后图像的灰度/色度值仍然在0 255范围内的对数变换，称为保带宽对数变换；满足这种要求的对数底数称为保带宽底数。设保带宽底数为
| $ {{\log }_{b}}256=255 $ | (3) |
则保带宽底数
定义相对反图像(或称补图像、负图像)的图像为正图像，对反图像再取反得到的也是正图像。定义对正图像的对数变换为正向对数变换；相应地，对反图像的对数变换称为反向对数变换。正向对数变换有增加图像灰度/色度谱低端(图像暗区)灰度/色度差、降低灰度/色度谱高端(图像亮区)灰度/色度差的功能；而反向对数变换功能与正向对数变换正好相反。类似的有正向保带宽对数变换和反向保带宽对数变换。
2 自适应双向保带宽对数变换
针对经典对数变换仅能提高图像暗区灰度级间差而不能同时提高图像亮区灰度级间差的缺陷，提出自适应双向保带宽对数变换。算法总体步骤如下：
1) 对输入的低照度图像进行标准化变换，得到标准化图像；
2) 计算标准化图像的平均亮度
3) 根据
4) 对自适应双向保带宽对数变换后的图像取整，输出最终增强后的图像。
步骤2) 中图像平均亮度
| $ {{L}_{\text{A}}}=\frac{1}{MN}\sum\limits_{x=0}^{M-1}{\sum\limits_{y=0}^{N-1}{g\left( x, y \right)}} $ | (4) |
式中，
| $ L_{\text{A}}^{\text{T}}=\frac{\sqrt{{{\left( L_{\text{A}}^{\text{R}} \right)}^{2}}+{{\left( L_{\text{A}}^{\text{G}} \right)}^{2}}+{{\left( L_{\text{A}}^{\text{B}} \right)}^{2}}}}{\sqrt{3}} $ | (5) |
高质量图像的平均亮度
3 实验结果与分析
3.1 主观评价
从29幅低照度图像中随机选取了两幅进行处理分析，处理结果如图 4和图 5所示。在图 4caps中，HE方法图像整体亮度得到提高，但处理后图像出现过增强和细节丢失现象，如图 4(c)右上角的云朵过亮，云朵丢失了亮暗的层次感；MSR处理的图像，如图 4(d)所示，帽子的边缘有明显的光晕伪影现象，且图像对比度降低；NPEA处理的图像对比度明显提高，但图像颜色失真，如图 4(e)所示4顶帽子的影子呈微红色、右上角的云朵泛黄；本文算法处理的图像，如图 4(f)所示，图像整体在对比度、亮度都有较大的提高，颜色也较自然，与参考图像图 4(a)非常逼近。
在图 5中，HE方法处理的图像过增强现象较严重，如图 5(c)所示，女士的脸、脖颈、手臂过亮，颜色失真；MSR处理的图像，如图 5(d)所示，图像整体出现灰化现象，对比度较低；而NPEA处理的图像对比度过高，如图 5(e)所示，女士本来黑色的长裙上出现了很多小亮点；本文算法处理的图像，如图 5(f)所示，女士的脸、黑色长裙等都较为自然，更符合人眼视觉，非常接近参考图像图 5(a)。
3.2 客观评价
为客观评价各算法对caps和woman图像的增强效果，现计算各增强图像与参考图像的PSNR和SSIM值，计算结果如表 1所示。从表 1可以看出，本文算法的PSNR值和SSIM值明显高于其余3种算法，这也从客观上印证了主观评价的结果，即本文算法处理的图像质量相对最好，最接近参考图像。
表 1
评价指标对比
Table 1
Comparison of evaluation indexes
为排除单个图像对算法评价指标的影响，现用多批箱线图(Box-plot)绘制各算法处理的所有低照度图像(共29幅)的PSNR、SSIM评价指标，如图 6、图 7所示，其中图 6为PSNR对比，图 7为SSIM对比。表 2为4种算法增强结果图像的PSNR和SSIM平均值对比。本文算法的PSNR和SSIM平均值分别为22.75 dB和0.86，明显高于其他3种算法，也即本文算法增强后的图像与参考图像失真误差最小。因此本文算法对低照度图像增强无论主观感觉还是客观评价均是优于其他3种增强算法。
表 2
PSNR和SSIM平均值对比
Table 2
Comparison of the mean values of PSNR and SSIM
| 评价指标 | 低照度图像 | HE | MSR | NPEA | 本文算法 |
| SNR/dB | 9.69 | 16.16 | 15.82 | 18.62 | 22.75 |
| SSIM | 0.37 | 0.58 | 0.62 | 0.78 | 0.86 |
为进一步衡量各算法运行效率，实验平台采用处理器为Intel Core i3-4160 3.60 GHz、内存为4 GB，运行Win7系统的PC机，实验采用的软件为Matlab2016a。实验分别对29幅低照度图像处理，4种算法的平均耗时如表 3所示。由表 3可以看出，本文算法与HE算法耗时接近，算法速度远远优于MSR、NPEA算法，可见本文算法具有较高的运行效率。
表 3
各算法平均耗时对比
Table 3
Average time-consuming comparison of each algorithm
| 平均耗时 | HE | MSR | NPEA | 本文算法 |
| |
0.043 | 11.28 | 11.58 | 0.074 |
4 结论
针对低照度图像亮度低、对比度差特点，本文通过分析传统对数变换仅能提高图像暗区对比度不能同时提高图像亮区对比度的缺陷，提出了自适应双向保带宽对数变换。方法的提出得益于前期对图像灰度/色度谱的深入分析。实验结果表明，本算法可有效增强低照度图像，并消除了Retinex光晕伪影现象，颜色更自然，更符合人眼视觉特性，算法简单，而且运算速度快，执行效率高。本文算法可广泛适用于背光或光照不均的低照度环境下的图像增强。
不过，本文算法的局限之处在于增强后的图像其对比度和亮度还有待提高的空间。因此，下一步工作是在保持自然的条件下，对增强后的图像进行指定平均亮度变换，以进一步增强对比度、增大亮度。另外，为提高算法的鲁棒性，对夜间视频监控增强还需做深入的测试和验证。
参考文献
-
[1] Jiang B J, Zhong M X. Improved histogram equalization algorithm in the image enhancement[J]. Laser & Infrared, 2014, 44(6): 702–706. [姜柏军, 钟明霞. 改进的直方图均衡化算法在图像增强中的应用[J]. 激光与红外, 2014, 44(6): 702–706. ] [DOI:10.3969/j.issn.1001-5078.2014.06.024]
-
[2] Sheet D, Garud H, Suveer A, et al. Brightness preserving dynamic fuzzy histogram equalization[J]. IEEE Transactions on Consumer Electronics, 2010, 56(4): 2475–2480. [DOI:10.1109/TCE.2010.5681130]
-
[3] Singh K, Kapoor R. Image enhancement using exposure based sub image histogram equalization[J]. Pattern Recognition Letters, 2014, 36(1): 10–14. [DOI:10.1016/j.patrec.2013.08.024]
-
[4] Jobson D J, Rahman Z, Woodell G A. Properties and performance of a center/surround retinex[J]. IEEE Transactions on Image Processing, 1997, 6(3): 451–462. [DOI:10.1109/83.557356]
-
[5] Rahman Z, Jobson D J, Woodell G A.Multi-scale retinex for color image enhancement[C]//Proceedings of International Conference on Image Processing.Lausanne:IEEE, 1996:1003-1006. [DOI:10.1109/ICIP.1996.560995]
-
[6] Lin H N, Shi Z W. Multi-scale retinex improvement for nighttime image enhancement[J]. Optik-International Journal for Light and Electron Optics, 2014, 125(24): 7143–7148. [DOI:10.1016/j.ijleo.2014.07.118]
-
[7] Wang S H, Zheng J, Hu H M, et al. Naturalness preserved enhancement algorithm for non-uniform illumination images[J]. IEEE Transactions on Image Processing, 2013, 22(9): 3538–3548. [DOI:10.1109/TIP.2013.2261309]
-
[8] Li H, Wu W, Yang X M, et al. Multispectral image enhancement based on Retinex by using structure extraction[J]. Acta Physica Sinica, 2016, 65(16): #160701. [李红, 吴炜, 杨晓敏, 等. 基于主特征提取的Retinex多谱段图像增强[J]. 物理学报, 2016, 65(16): #160701. ] [DOI:10.7498/aps.65.160701]
-
[9] Xiao L M, Li C, Wu Z Z, et al. An enhancement method for X-ray image via fuzzy noise removal and homomorphic filtering[J]. Neurocomputing, 2016, 195: 56–64. [DOI:10.1016/j.neucom.2015.08.113]
-
[10] Tian X P, Cheng X, Wu C M, et al. Color image enhancement method based on homomorphic filtering[J]. Journal of Xi'an University of Posts and Telecommunications, 2015, 20(6): 51–55. [田小平, 程新, 吴成茂, 等. 基于同态滤波的彩色图像增强[J]. 西安邮电大学学报, 2015, 20(6): 51–55. ] [DOI:10.13682/j.issn.2095-6533.2015.06.011]
-
[11] Gonzalez R C, Woods R E.Digital Image Processing[M].Ruan Q Q, Ruan Y Z, trans.3rd ed.Beijing:Publishing House of Electronics Industry, 2011:64-66. [冈萨雷斯, 伍兹. 数字图像处理[M]. 阮秋琦, 阮宇智, 译. 3版. 北京: 电子工业出版社, 2011: 64-66.]
-
[12] Chen Y, Li Y, Lv X F, et al. Application of normalized transformation in invisible video image reproduction[J]. Semiconductor Optoelectronics, 2013, 34(2): 307–310, 315. [陈勇, 李愿, 吕霞付, 等. 标准化变换在视频隐形图像重现中的应用[J]. 半导体光电, 2013, 34(2): 307–310, 315. ]
-
[13] Xie Z X, Wang Z F, Liu Y H, et al. The theory of gradually flattening gray spectrum[J]. Chinese Journal of Medical Physics, 2006, 23(6): 405–407, 426. [谢正祥, 王志芳, 刘燕欢, 等. 灰度谱分级平坦化理论[J]. 中国医学物理学杂志, 2006, 23(6): 405–407, 426. ] [DOI:10.3969/j.issn.1005-202X.2006.06.003]
-
[14] Xie Z X, Wang Y, Peng Z Y, et al. The image hiding and mining technique based on the Zadeh-X transformation[J]. Chinese Journal of Medical Physics, 2007, 24(1): 9–11. [谢正祥, 王颖, 彭子苡, 等. 基于Zadeh-X变换的图像隐藏和挖掘技术[J]. 中国医学物理学杂志, 2007, 24(1): 9–11. ] [DOI:10.3969/j.issn.1005-202X.2007.01.004]
-
[15] Xiong X L, Wang Z F, Yang J Y, et al.Standard image generation method:China, CN102800062A[P].2012-11-28. [熊兴良, 王志芳, 杨佳义, 等. 标准化图像的生成方法: 中国, CN102800062A[P]. 2012-11-28.]
-
[16] Xie Z X, Wang Z F, Lv X F.Single image quality assessment and creation of the best quality image[C]//Proceedings of the 2nd International Conference on Information Science and Engineering.Hangzhou:IEEE, 2010:4022-4025.[DOI:10.1109/ICISE.2010.5688646]
-
[17] Wang Z, Bovik A C, Sheikh H R, et al. Image quality assessment:from error visibility to structural similarity[J]. IEEE Transactions on Image Processing, 2004, 13(4): 600–612. [DOI:10.1109/TIP.2003.819861]
-
[18] Sheikh H R, Wang Z, Cormack L, et al.LIVE Image quality assessment database release 2[DB/OL].[2017-02-18].http://live.ece.utexas.edu/research/quality/subjective.htm.
