Icamme managed brightness and contrast enhancement using adapted histogram equalization

Available online at www.sciencedirect.com
Procedia Materials Science (2014) 000–000
www.elsevier.com/locate/procedia
2211-8128 © 2014 The Authors. Published by Elsevier Ltd.
Selection and peer-review under responsibility of Organizing Committee of AMME 2014.
International Conference on Advances in Manufacturing and Materials Engineering,
AMME 2014
Managed Brightness and Contrast Enhancement using Adapted
Histogram Equalization
K. Santhi a*
, RSD. Wahida Banub
, R. Dhanasekaranc
a*, c
Assistant Professor, K. S. Rangasamy College of Technology,
Tiruchengode – 637215, Tamilnadu, India
b
Principal, Government College of Engineering, Salem – 636011, Tamilnadu, India
Abstract
Modified enhancement managed contrast using adjusted histogram is developed to minimize the problems of over enhancement,
saturation artifacts and change in mean brightness with conventional histogram equalization. The input image’s histogram is first
divided into four sub-histograms based on its median. A clipping process based on the input image mean is applied. Then each
partitioned histogram is equalized independently. A contrast enhancement rate is devised in order to achieve the varying contrast
for output images. The proposed algorithm is proved to produce better enhanced images than the contemporary techniques in
terms of contrast per pixel and structural similarity index.
© 2014 The Authors. Published by Elsevier Ltd.
Selection and peer-review under responsibility of Organizing Committee of AMME 2014.
Keywords: Contrast – Brightness - Enhancement - Histogram equalization.
1. Introduction
Due to portability and easy implementation, the digital camera has become an additional feature embedded in
cell phones by many manufacturers. However, the qualities of the images captured using a cell phone camera is
usually poor as a result of low contrast by Siddiqui H., (2008). In a dark room or during night time, the lack in
natural light-sources leads to poor and lowly contrasted images. To overcome this problem, recently LEDs are used
for dark environments. However, the lighting from LEDs of capturing devices is insufficient to brighten the course
between the image and the images capturing device which are at a longer distance. As a result, the images captured
produce annoying artifacts for low contrast by Gamadia, M. et al. (2007).
Histogram Equalization (HE) is a popular technique for enhancing image contrast, is to map the gray levels
based on the probability distribution of the input image gray levels. HE flattens and stretches the dynamic range of

2 Author name / Procedia Materials Science 00 (2014) 000–000
an image histogram and gives an overall contrast improvement by Gonzalez, R. C., and Woods, R. E., (2002). HE
has been used in all fields like Medical, Radar, Satellite and Microstructure image processing. The HE is available
in most image processing packages such as Adobe Photoshop by Weichselbaum, M., et al. (2005), National
Institutes of Health Imageby Rasband, W. S., and Bright, D. S., (1995) and Lispix by Wong, W. K., (1997). But the
enhanced image tends to have unnatural enhancement and intensity saturation artifacts due to the error in brightness
because of mean-shifting due to by HE Kim, T., and Paik, J., (2008). In the recent years, many researchers have
proposed several useful algorithms to solve the problems of HE by Sim, K. S., et al. (2007), Chen, S. D., and Ramli,
A. R., (2003) and Kim, Y. T., (1997). Generally these contrast enhancement using HE can be divided into two
partitioned histogram equalization or dynamic partitioned histogram equalization (DPHE). One important thing of
DPHE is assigned to a new enhanced dynamic range instead of using the original dynamic rangeby Chee Hee Ooi,
(2010). Its partition is based on mean or median values. The mean preserving bi-histogram equalization with
neighborhood metrics has been proposed to overcome the saturation effect. First, an original image histogram is
created using the distinction neighborhood metric and it is divided into 2040 sub bins. Second, the histogram is
separated into two sub-histograms, which are equalized independently.
Recently, Range Limited Bi-Histogram Equalization for Image Contrast Enhancement (RLBHE) has been
developed by Chao Zuo et al. (2012). This method divides the input histogram into two independent sub-histograms
by a threshold which is used to separate the object from the background that minimizes the intra class variance.
Then the range of the output image is calculated to yield minimum absolute mean brightness error between the
original and equalized image.
This article presents a Managed Brightness and Contrast Enhancement using Adapted Histogram (CCMHE). The
main objective of contrast enhancement is to process an image in order to get a better result that is more suitable than
that of the original image. Therefore the proposed method divides the histogram into four parts based on the median
brightness. The enhancement rate parameter plays an important role in CCMHE. Simulation results showed that this
proposed algorithm is more suitable to enhance the contrast and to maintain the brightness. Structural similarity
index and Contrast per pixel are used to analyze the image quality.
2. Controlled Contrast Modified Histogram Equalization (CCMHE)
Controlled contrast using modified histogram (CCMHE) which is proposed in this paper consists of six steps.
The details of each step are described in the following subsections.
2.1. Average Difference Metrics
Histogram of original image is divided into two sub-histograms. For separating point, median values are used to
divide the two sub-histograms into two smaller associate-histograms and totally four sub-histograms are obtained.
In the input histogram, the minimum and maximum intensity values (0,255) are used to set as the separating points.
Each and every separating point can be calculated by the equations (1-4),
𝑞1
= 0.2 {𝑅 𝑤𝑖𝑑𝑡𝑕 × 𝑅 𝑕𝑒𝑖𝑔𝑕𝑡} (1)
𝑞2
= 0.4 {𝑅 𝑤𝑖𝑑𝑡𝑕 × 𝑅 𝑕𝑒𝑖𝑔𝑕𝑡} (2)
𝑞3
= 0.6 {𝑅 𝑤𝑖𝑑𝑡𝑕 × 𝑅 𝑕𝑒𝑖𝑔𝑕𝑡} (3)
𝑞4
= 0.8 {𝑅 𝑤𝑖𝑑𝑡𝑕 × 𝑅 𝑕𝑒𝑖𝑔𝑕𝑡} (4)
where q1, q2, q3 and q4 are the total number of pixels in the input image histogram intensities set to 0.2, 0.4, 0.6
and 0.8 respectively. R width and R height represent the width and the height of the input image.

Author name / Procedia Materials Science 00 (2014) 000–000 3
2.2. Clipping Process
In order to overcome the unnatural and over enhancement of processed image, clipping process is used. For this
automatic clipping process, the self adaptive plateau HE for the infrared image contrast enhancement is proposed.
This algorithm may fail to detect local peak detection for classical images. So, a modified-SPAHE by Wong, W. K.,
(1997) was introduced to locate median value of the non empty bins as the clipping threshold Tc. In this proposed
work, to minimize computational complexity, average numbers of intensity values were taken for implementation.
2.3. New Gray Point Array Distribution
Partition based histogram equalization methods perform the enhancement process in each sub-histogram which
are between two separating points. Thus the sub-histograms may not ensure the balance space in each sub-histogram
for sufficient contrast enhancement because contrast enhancement obtained in a narrow stretching space is less
significant. This phenomenon particularly occurs when the side of the sub-histogram is narrow. Consequently, the
processed image tends to suffer from loss of image details and intensity saturation artifacts by Siddiqui H. (2008).
To balance the enhancement space for each sub-histogram, a novel gray level dynamic range method is employed in
this proposed method.
𝐷𝑦𝑛𝑎𝑚𝑖𝑐 𝑟𝑎𝑛𝑔𝑒 =
𝐺𝑟𝑎𝑦 𝑙𝑒𝑣𝑒𝑙 𝑠𝑝𝑎𝑛𝑠
𝑇𝑜𝑡𝑎𝑙 𝑛𝑜 𝑜𝑓 𝑝𝑖𝑥𝑒𝑙𝑠
(5)
𝑆𝑝𝑎𝑛 𝑟
= 𝑞 𝑟+1
− 𝑞 𝑟
(6)
𝑓𝑎𝑐𝑡𝑜𝑟 𝑟
= 𝑠𝑝𝑎𝑛 𝑟
× (𝑙𝑜𝑔10
𝑄 𝑟) δ
(7)
where spanr in (10) is the dynamic gray level used by rth
sub-histogram in the input image, qr is the rth
separating
point, Qr is the total number of pixels in rth
sub histogram and δ is the amount of emphasis, which is the adjusted
parameter. In the input image, its dynamic range of the gray levels is ‘0’ i.e. δ=0. In this moment, the image details
will be at a loss. Then the value of δ is important and it is discussed in next subdivision.
𝑟𝑎𝑛𝑔𝑒 𝑟
=
𝑓𝑎𝑐𝑡𝑜𝑟 𝑟
𝑛
𝑘=1 𝑓𝑎𝑐𝑡𝑜𝑟 𝑘
∗ (𝐿 − 1) (8)
where k = 0,1…,n (12). In the rth
sub-histogram the new dynamic range is allocated from [rmin, rmax], where
𝑟 𝑚𝑖𝑛 = 𝑟 − 1 𝑚𝑎𝑥 − 1 (9)
𝑟 𝑚𝑎𝑥 = 𝑟 𝑚𝑖𝑛 + 𝑟𝑎𝑛𝑔𝑒 (10)
The rmin represents minimum intensity value of new dynamic range. ranger states the gray level range for ith
sub-histogram in the output image.
2.4. Histogram Equalization
After obtaining the dynamic range as in equation (9), the new ranges are found in all sub histograms. At last,
each sub-histogram is equalized independently. The rth
histogram is assigned at a gray level range in [rmin,rmax]. Then
the output histogram equalization Y is calculated using following transfer function,
𝑌 = (𝑟 𝑚𝑖𝑛 − 𝑟𝑚𝑎𝑥 ) ∗ 𝐶𝐷𝐹(𝑋𝑘 ) + 𝑟 𝑚𝑖𝑛 (11)

where CDF(Xk) – cumulative density function in sub histogram, rmin and rmax are minimum and maximum
intensities in the output range, respectively.
2.5. Enhancement rate controlling process
In QDHE, in order to minimize the computational complexity, amount of emphasis (enhancement rate parameter)
is removed. So, the output of image is unnatural and also the enhancement is not up to the level. In this approach, δ
is the only parameter that needs to be adjusted. It can be adjusted by using span value of each sub-histogram in the
output. Suppose, the value of δ = 0, it means that the dynamic range of the gray levels of the input is less. Otherwise
δ value is set by user.
2.6. Adjusting the level of enhancement rate
HE does not provide provision for adjusting the level of enhancement. But, the proposed work gives a provision
to have a control over the level of enhancement. The main objective of this method is to find a modified histogram ђ
that is closer to uniform histogram f and to achieve natural looking of enhanced images in order to adjust the
enhancement level.
Ђ =
𝑕 𝑘+ 𝛿 𝑓
1+ 𝛿
= (
1
1 + 𝛿
)𝑕 𝑘 + (
𝛿
1+ 𝛿
)𝑓 (12)
From the modified histogram ђ, weighted average of hk and f can be obtained, by changing the value of the level
of enhancement by Sundaram, M. (2011). The enhancement parameter δ varies from 0 to α. suppose δ = infinity, it
converges to preserve the original. So, various levels of contrast enhancement can be achieved by varying the value
of δ [13]. The level of contrast enhancement should be adjusted depending on the image’s contrast. In our proposed
algorithm δ value is taken as 5 for standard database images.
3. Measurement Tools to Assess Image Quality
3.1. Structural Similarity Index
Natural image signals are highly structured. Their pixels exhibit strong dependencies, especially when they are
spatially proximate and these dependencies carry important information about the structure of the objects in the
visual scene.
𝑆𝑆𝐼 (𝑥, 𝑦) =
(2𝜇 𝑥 𝜇 𝑦 +𝐶1)(2𝜎 𝑥𝑦 +𝐶2)
(𝜇 𝑥
2+𝜇 𝑦
2 +𝐶1)(𝜎 𝑥
2+𝜎 𝑦
2+𝐶2)
(13)
where μx is the mean intensity of the input image x, μy is the mean intensity of the output image y, σx and σy are
the standard deviation of the input image x and output image y respectively, σxy is the square root of covariance of
images x and y and C1 and C2 are constants.
3.2. Contrast-per-pixel
Contrast-per-pixel (CPP) measures the average intensity difference between a pixel and its adjacent pixels. This
value shows the local contrast of the image by Sangee, N., et al. (2010).
𝐶 =
𝑁−1
𝑖=0
𝑀−1
𝑗=0 (
(𝑚 ,𝑛)∈𝑅3
(𝑖,𝑗 ) |𝛾(𝑖,𝑗 )−𝛾(𝑚,𝑛)|)
𝑀∗𝑁∗8
(14)

4. Result and Discussion
The performance of the controlled contrast modified histogram equalization technique (CCMHE) is tested on
more than 60 images. These images are subjected to contrast controlled modified histogram equalization process.
The level of enhancement is varied from 0 to 1000. In this work is focused by 0 to 5.
 Simulation results are obtained using a variety of original images by three evaluations:
 Quantitative assessment for all test and real time images is done by using different well recognized metrics.
 Qualitative evaluation of enhancement results is done by seven contemporary methods.
 Time complexity is analyzed for all existing contrast enhancement method.
To compare this CCMHE algorithm, similar images are involved with the existing methods HE, BBHE, HS,
RMSHE (recursive level 2), MMBEBHE, QDHE, BHENM and RLBHE. The performance of this method is
checked qualitatively and quantitatively in terms of human perception, contrast per pixel and SSIM.
Fig. 1Original SEM image and its histogram equalization
Fig. 1 shows the image of dimple fracture of steel (gear) tooth. This image is captured from SEM model JSM –
6360, PSG Technology, Coimbatore, India. The scanning electron microscope (SEM) is used to focus at the surface
of the gear component. The beam of high-energy electrons is produced a variety of signals. This magnification is
500x. The algorithms are made to deal with a SEM image of a steel component, taken in an industry.

Fig. 2 SEM image (a) enhanced results of HE (b) BBHE (c)HS (d) RMSHE (r=2) (e) MMBEBHE (f) QDHE (g) BHENM (h) CCMHE
The Fig. 1 corresponds to the original SEM image and the histogram associated with it. Such an image is
provided as the input for the algorithms discussed here and the outputs are evaluated. Fig. 2(a - h) are the resultant
images for HE, BHE, MMBEBHE, Histogram specification, RMSHE, RLBHE, BHENM, QDHE and the proposed
routine. From the histogram part Fig. 2(a), it can be clearly understood that the output is not the exact reproduction
of the input.
The effects of BHE, MMBEBHE and Histogram specification are also similar to that of the HE method and so
they are also not the intended results. The RMSHE technique provides an output with a bit of oscillation when
compared with the input SEM image as indicated in the Fig. 2(d). In addition to them, the results of MMBEBHE,
Histogram specification and RMSHE can be clubbed together as they have slight oscillations only in a smaller
section of the histogram. Their outputs are somewhat better than that of the previous case but not in a proper way.
From the output of this proposed method CCMHE as shown in Fig 2(h), the complete analysis can be done at ease.
It is so because, the algorithm proposed is highly effective than that of the methods which are discussed. In order to
show the performance of our proposed method, use the four different metrics to assess the quality of the processed
images. The results were given and Table 1 shows that performance measure values from different techniques for
SEM image .JPG. It clearly shows the proposed method contrast per pixel values produced by CCMHE is high. For
all images SSI values are higher than the contemporary methods, because this algorithm increases the quality of
images.
Table 1 Performance Measure values from different techniques for SEM image .JPG
Measures HE BBHE HS RMS HE MMBEBHE QDHE BHE NM RLB HE Proposed CCMHE
CPP 21.59 24.27 36.74 36.74 23.61 19.47 25.48 25.48 17.145
SSI 0.763 0.765 0.7866 0.786 0.8539 0.6164 0.741 0.742 0.8601
5. Conclusion
In this paper, controlled contrast using adjusted histogram is proposed. The input image’s histogram is first
divided into four sub-histograms based on its median. A clipping process based on the input image mean is applied.
Then each partitioned histogram is equalized independently. The proposed algorithm is compared to seven
contemporary methods. Based on the experimental results, this method enhances the contrast more naturally. In
order to adjust the control parameter, the effect of output is discussed. The enhancement rate is 5 for getting visually
pleasing output. Furthermore, by quantitative analysis can be obtained through different parameters namely
Structural Similarity Index (SSI) and Contrast Per Pixel (CPP). From the simulation results, it is clear that the
proposed method is suitable for improve the quality of low contrast images as well as real time images. In future,
this algorithm can be extended to color images for different industrial application of better visualization.
References
Chao Zuo, Qian Chen, Xiubao Sui, et al. 2012. Range Limited Bi-Histogram Equalization for image contrast enhancement. Optik, 1-5.

Chee Hee Ooi, Nor Ashidi Mat Isa, et al. 2010. Quadrants Dynamic Histogram Equalization for Contrast Enhancement. IEEE Trans. Consumer
Electronics, 56, 2552-2559.
Chen, S. D., & Ramli, A. R., 2003. Minimum mean brightness error bi-histogram equalization in contrast enhancement. IEEE Trans. Consumer
Chen, S. D., & Ramli, A. R., et al. 2003. Contrast enhancement using recursive mean separate histogram equalization for scalable brightness
preservation. IEEE Trans. Consumer Electronics, 49, 1301-1309.
Gamadia, M., Kehtarnavaz, N., Hoffman, K. R., et al. 2007. Low-Light Auto-Focus Enhancement for Digital and Cell-Phone Camera Image
Pipelines. IEEE Trans. Consumer Electronics, 53, 249-257.
Gonzalez, R. C., & Woods, R. E., 2002. Digital Image Processing. 2nd ed., Prentice Hall.
Kim, T., & Paik, J., 2008. Adaptive contrast enhancement using gain-controllable clipped histogram equalization. IEEE Trans. Consumer
Kim, Y. T., 1997. Contrast enhancement using brightness preserving bi-histogram equalization. IEEE Trans. Consumer Electronics, 43, 1-8.
Rasband, W. S., & Bright, D. S., 1995. NIH Image: a public domain image processing program for the Macintosh. Microbeam Analysis Society,
4, 137–149.
Sangee, N., Sangee, A., Choi, H. K., et al. 2010. Image contrast enhancement using bi-histogram equalization with neighborhood metrics. IEEE
Trans. Consumer Electronics, 56(4), 2552-2559.
Siddiqui H., Bouman, C. A., et al. 2008. Hierarchical Color Correction for Camera Cell Phone Images. IEEE Trans. Image Processing, 17, 2138-
2155.
Sim, K. S., Tso, C. P., Tan, Y. Y., et al. 2007. Recursive sub-image histogram equalization applied to gray-scale images. Pattern Recognition
Letters, 28, 1209-1221.
Sundaram, M., Ramar, K., Arumugam, N., Prabin, G., et al. 2011. Histogram Modified Local Contrast Enhancement for mammogram images.
Applied Soft Computing, 11, 5809-5816.
Wan, Y., Chen, Q., Zhang, B. M., 1999. Image enhancement based on equal area dualistic sub-image histogram equalization method. IEEE Trans.
Consumer Electronics, 45, 68-75.
Weichselbaum, M., Sparrow, M. P., Hamilton, E. J., Thompson, P. J.,Knight, D. A., 2005. A confocal microscopic study of solitary pulmonary
neuroendocrine cells in human airway epithelium. Respiratory Research, 6, 1–11.
Wong, W. K., 1997. Control of specimen charging in scanning electron microscopy. Ph.D. Thesis, National University of Singapore, Singapore.

Icamme managed brightness and contrast enhancement using adapted histogram equalization

More Related Content

What's hot (19)

Similar to Icamme managed brightness and contrast enhancement using adapted histogram equalization (20)

Recently uploaded (20)

Icamme managed brightness and contrast enhancement using adapted histogram equalization