A comparative study of histogram equalization based image enhancement techniques for brightness preservation and contrast enhancement

Signal & Image Processing : An International Journal (SIPIJ) Vol.4, No.5, October 2013

A COMPARATIVE STUDY OF HISTOGRAM
EQUALIZATION BASED IMAGE
ENHANCEMENT TECHNIQUES FOR
BRIGHTNESS PRESERVATION AND CONTRAST
ENHANCEMENT
Omprakash Patel, Yogendra P. S. Maravi and Sanjeev Sharma
School of Information Technology, RGPV, Bhopal, Madhya Pradesh, India

ABSTRACT
Histogram Equalization is a contrast enhancement technique in the image processing which uses the
histogram of image. However histogram equalization is not the best method for contrast enhancement
because the mean brightness of the output image is significantly different from the input image. There are
several extensions of histogram equalization has been proposed to overcome the brightness preservation
challenge. Contrast enhancement using brightness preserving bi-histogram equalization (BBHE) and
Dualistic sub image histogram equalization (DSIHE) which divides the image histogram into two parts
based on the input mean and median respectively then equalizes each sub histogram independently. This
paper provides review of different popular histogram equalization techniques and experimental study based
on the absolute mean brightness error (AMBE), peak signal to noise ratio (PSNR), Structure similarity index
(SSI) and Entropy.

KEYWORDS
Histogram Equalization, Contrast Enhancement, Brightness Preservation, Absolute Mean Brightness Error,
Peak Signal to Noise Ratio, Structure Similarity Index.

1. INTRODUCTION
Contrast enhancement is an important area in the field of digital image processing for human
visual perception and computer vision. It is extensively used for medical image processing and as
a preprocessing step in speech recognition, texture synthesis, and many other image/video
processing applications [1] [2] [3]. There are many different methods have been developed for
image contrast enhancement [4]-[22]. Here we discussed some popular image contrast
enhancement method for brightness preservation.
Contrast Enhancement is a specific characteristics enhancement in the image enhancement
processing. In which histogram equalization (HE) is the one of the popular method of image
contrast enhancement [4]. The histogram of the discrete gray-level image represent the frequency
DOI : 10.5121/sipij.2013.4502

11


of occurrence of all gray-levels in the image [5]. Histogram equalization well distributes the
pixels intensity over the full intensity range. But HE has "mean-shift" problem [6], it shifts the
mean intensity value to the middle gray level of the intensity range. So this technique is not useful
where brightness preservation is required. To overcome this problem BBHE (Brightness
Preserving by Histogram Equalization) has been proposed by Yeong-Taeg Kim [6] which
overcome the mean shift problem. BBHE divides the input image histogram into two parts based
on the mean value. Then histogram equalization is applied to the both separated parts with the
new intensity range, first minimum gray level to the mean value and for second mean value to the
maximum gray level.
Later, equal area dualistic sub-image histogram equalization (DSIHE) introduced by Yu Wan,
Qian Chen and Bio-Min Zang [7] and claimed that the performance of DSIHE is better than
BBHE in terms of brightness preservation and entropy. DSIHE uses the same concept as BBHE,
it divides the input image histogram into two parts based on cumulative probability density equal
to 0.5 instead of mean value as in BBHE. It divides the image in two parts with equal size which
contains equal number of pixel.
The extension of BBHE is RMSHE which perform recursively separation of input image
histogram based on mean value of input histogram to the satisfactory number of recursion levels.
Each new sub-histogram is separated based on respective mean. As the number of recursion level
increases the mean brightness of output image is closer to the input image mean brightness. The
recursive nature of RMSHE implies scalable brightness preservation which is very useful in
consumer electronic products [8]. A new method RSIHE is a novel extension of DSIHE and used
some characteristics of RMSHE. The RSIHE recursively separated the input image histogram
based on the cumulative distribution function (cdf) equal to 0.5. The RSIHE separates each new
sub histogram recursively to the specified satisfactory recursion level [10].
Chen and Ramil proposes Minimum Mean Brightness Error Bi-Histogram Equalization
(MMBEBHE), which is an extension of BBHE which offers maximum brightness preservation.
MMBEBHE method proposes to separate histogram into two sub-parts using threshold level
which is responsible for minimum absolute mean brightness error (AMBE) [9].
Mary Kim and Min Gyo Chung proposed another new method RSWHE (Recursively Separated
and Weighted Histogram Equalization) for brightness preservation and contrast enhancement.
The RSWHE offers better results in comparison to previous methods in all aspects. The RSWHE
segments the input histogram into two or more sub histogram recursively based on the mean or
median as used in RMSHE and RSIHE, then apply the weighting process based on normalized
power low function to modify the each sub-histogram. And then perform histogram equalization
to each weighted output sub-histogram independently [12].

A. HISTOGRAM EQUALIZATIONHistogram equalization is contrast enhancement technique in a spatial domain in image
processing using histogram of image. Histogram equalization usually increases the global contrast
of the processing image. This method is useful for the images which are bright or dark.

12


Implementation
Consider the discrete grayscale input Image X = x(i,j), with the L discrete levels, where x(i,j)
represents the intensity levels of the image at the spatial domain (i,j) Let histogram of Image X is
H(X). Now the probability density function pdf(X) can be defined as.
݊௞
ൗܰ
ሻ
Where, 0 ൑ ݇ ൑ ሺ‫ ܮ‬െ 1ሻ

‫݂݀݌‬ሺܺ௞ ሻ‫݌ ݎ݋‬ሺܺ௞ ሻ ൌ

(1)

L is the total number of gray levels in the image,
N is the Total number of pixels in the image,
nk is the total number of pixels with the same intensity level k.
From the pdf(X) (1) the cumulative distribution function cdf(Xi) is defined as݂ܿ݀ሺܺ௜ ሻ‫ܿ ݎ݋‬ሺܺ௜ ሻ ൌ ∑௞ ‫݌‬ሺܺ௜ ሻ
௜ୀ଴

(2)

Note that ݂ܿ݀ሺܺ௅ିଵ ሻ ൌ 1 from eq. (1) and (2).

(a)

(c)

(b)

(d)

Fig. 1. (a) shows input image having low contrast, (b) shows histogram of input image, (c) shows
histogram equalized image and (d) shows histogram of processed image

13


Histogram equalization is a scheme that maps the input image into the entire dynamic range
[ܺ଴ , ܺ௅ିଵ ] by using the cumulative distribution function as a transform function. Let's define the
transform function ݂ሺܺሻ using cumulative distribution function ݂ܿ݀ሺܺ௜ ሻ as݂ሺܺሻ ൌ ܺ଴ + ሺܺ௅ିଵ െ ܺ଴ ሻ × ݂ܿ݀ሺܺ௜ ሻ
Then the output image of histogram equalization, ܻ ൌ ‫ݕ‬ሺ݅, ݆ሻ can be expressed as

(3)

ܻ ൌ ݂ሺܺሻ

(4)

ܻ ൌ ൛݂൫‫ݔ‬ሺ݅, ݆ሻ൯ห ∀ ‫ݔ‬ሺ݅, ݆ሻ ∈ ܺ}

(5)

The above describe the histogram equalization on gray scale image. However it can also be used
on color image by applying the same method separately to the Red, Green and Blue Component
of the RGB color image.

B. Brightness Preserving Bi-Histogram Equalization (BBHE)The BBHE firstly decomposes an input image histogram H(X) into two sub-images based on the
mean of the input image. Now let input image ܺis decomposed into two sub-images XL and XU by
using the input mean Xm, where ܺ௠ ∈ {ܺ଴ , ܺଵ , ܺଶ … ܺ௅ିଵ }.
ܺ ൌ ܺ௅ ∪ ܺ௎
where

and

(6)

ܺ௅ ൌ {‫ݔ‬ሺ݅, ݆ሻ| ‫ݔ‬ሺ݅, ݆ሻ ൑ ܺ௠ ∀ ‫ݔ‬ሺ݅, ݆ሻ ∈ ܺ}
ܺ௎ ൌ {‫ݔ‬ሺ݅, ݆ሻ| ‫ݔ‬ሺ݅, ݆ሻ > ܺ௠ ∀ ‫ݔ‬ሺ݅, ݆ሻ ∈ ܺ}

And mean of the input image is calculated by,
ܺ௠ ൌ

൫∑ಽషభ ௜.௣ሺ௑೔ ሻ൯
೔సబ
൫∑ಽషభ ௣ሺ௑೔ ሻ൯
೔సబ

(7)

Note that the sub-image XL is composed of {ܺ଴ , ܺଵ , ܺଶ … ܺ௠ } and sub-image XU is composed of
{ܺ௠ାଵ , ܺ௠ାଶ , ܺ௠ାଷ … ܺ௅ିଵ }
Next, define the respective probability density function (pdf) of the sub-image XL and XU as‫݂݀݌‬௅ ሺܺ௞ ሻ‫݌ ݎ݋‬௅ ሺܺ௞ ሻ ൌ
‫ݓ‬ℎ݁‫ ݇ ,݁ݎ‬ൌ 0,1,2 … ݉.
and

݊௞
ൗܰ
௅

(8)

14


݊௞
ൗܰ
௎
‫ݓ‬ℎ݁‫ ݇ ,݁ݎ‬ൌ ݉ + 1, ݉ + 2, … ‫ ܮ‬െ 1.
‫݂݀݌‬௎ ሺܺ௞ ሻ‫݌ ݎ݋‬௎ ሺܺ௞ ሻ ൌ

(9)

Where NL and NU represents the total number or pixels in the XL and XU respectively. The
respective cumulative distribution function (cdf) is݂ܿ݀௅ ሺܺ௜ ሻ ൌ ∑௠ ܲ௅ ሺܺ௜ ሻ
௜ୀ଴

(10)

݂ܿ݀௅ ሺܺ௜ ሻ ൌ ∑௅ିଵ ܲ௎ ሺܺ௜ ሻ
௜ୀ௠ାଵ

(11)

and

Where ܺ௜ ൌ ‫ .ݔ‬note that ݂ܿ݀௅ ሺܺ௠ ሻ ൌ 1 and ݂ܿ݀௅ ሺܺ௅ିଵ ሻ ൌ 1 by definition.
Now defining transformation function as defined in histogram equalization for each sub image݂௅ ሺܺ௅ ሻ ൌ ܺ଴ + ሺܺ௠ െ ܺ଴ ሻ݂ܿ݀௅ ሺܺ௜ ሻ
and

(12)

݂௎ ሺܺ௎ ሻ ൌ ܺ௠ାଵ + ሺܺ௅ିଵ െ ܺ௠ାଵ ሻ݂ܿ݀௎ ሺܺ௜ ሻ

(13)

Based on these transformation functions, sub images are equalized independently and the output
of BBHE is composition of equalized sub images. Now the output of BBHE, Y can be expressed
asܻ ൌ {ܻሺ݅, ݆ሻ}
ൌ ݂௅ ሺܺ௅ ሻ ∪ ݂௎ ሺܺ௎ ሻ
where
݂௅ ሺܺ௅ ሻ ൌ ݂௅ ሺܺሺ݅, ݆ሻሻ|∀ܺሺ݅, ݆ሻ ∈ ܺ௅

(14)

(15)

and
݂௎ ሺܺ௎ ሻ ൌ ݂௎ ሺܺሺ݅, ݆ሻሻ|∀ܺሺ݅, ݆ሻ ∈ ܺ௎

(16)

݂௅ ሺܺ௅ ሻ Equalize the sub-image ܺ௅ over the range [ܺ଴ , ܺ௠ ] whereas ݂௎ ሺܺ௎ ሻ equalize the
subimage ܺ௎ over the range [ܺ௠ାଵ , ܺ௅ିଵ ].

C. Equal Area Dualistic Sub-Image Histogram Equalization (DSIHE):Equal Area Dualistic Sub-Image Histogram Equalization (DSIHE) follows the same idea as in the
BBHE, which decomposes the original image histogram into two sub histogram based on the
mean value. DSIHE method decomposes the image based on the gray level with a cumulative
distribution value equals to 0.5 instead of the men as in BBHE method. DSIHE method
15


decomposes the image aiming at the maximization of the Shannon’s entropy of the output image.
For such aim the input image is decomposed into two sub-images, being one dark and one bright,
and then applies the HE on the decomposed sub-images. Then compose the resultant histogram
equalized sub-image into one image to produce the DSIHE output image [7]. Assuming ܺ஽ is the
median value which divides input image into two parts ܺ௅ and ܺ௎ then ܺ௅ and ܺ௎ is be defined
asܺ ൌ ܺ௅ ∪ ܺ௎
(17)
where
ܺ௅ ൌ {‫ݔ‬ሺ݅, ݆ሻ| ‫ݔ‬ሺ݅, ݆ሻ ൑ ܺ஽ ∀ ‫ݔ‬ሺ݅, ݆ሻ ∈ ܺ}
and
ܺ௎ ൌ {‫ݔ‬ሺ݅, ݆ሻ| ‫ݔ‬ሺ݅, ݆ሻ > ܺ஽ ∀ ‫ݔ‬ሺ݅, ݆ሻ ∈ ܺ}
and
ܺ஽ ൌ arg ‫݊݅ܯ‬଴ஸ௞ஸ௅ିଵ ቚ݂ܿ݀ሺܺ௞ ሻ െ

௖ௗ௙ሺ௑బ ሻା௖ௗ௙ሺ௑ಽషభ ሻ
ቚ
ଶ

(18)

Where sub-image ܺ௅ is composed by the {ܺ଴ , ܺଵ , … . . , ܺ௘ିଵ }, and sub-image ܺ௎ is composed by
the {ܺ௘ , ܺ௘ାଵ , … . . , ܺ௅ିଵ }. Now let’s probability density function for sub-images ܺ௅ ܽ݊݀ ܺ௎
respectively are‫݂݀݌‬௅ ሺܺ௞ ሻ‫݌ ݎ݋‬௅ ሺܺ௞ ሻ ൌ

݊௞
ൗܰ , ‫ݓ‬ℎ݁‫ ݇ ݁ݎ‬ൌ 0,1,2 … ‫.ܦ‬
௅

(19)

‫݂݀݌‬௎ ሺܺ௞ ሻ‫݌ ݎ݋‬௎ ሺܺ௞ ሻ ൌ

݊௞
ൗܰ , ‫ݓ‬ℎ݁‫ ݇ ݁ݎ‬ൌ ‫ ܮ … 3 + ܦ ,2 + ܦ ,1 + ܦ‬െ 1.
௎

(20)

and

Cumulative distribution function for both sub-images ܺ௅ ܽ݊݀ܺ௎
݂ܿ݀௅ ሺܺ௜ ሻ ൌ ∑஽ିଵ ܲ௅ ሺܺ௜ ሻ
௜ୀ଴
and

(21)

݂ܿ݀௅ ሺܺ௜ ሻ ൌ ∑௅ିଵ ܲ௎ ሺܺ௜ ሻ
௜ୀ஽

(22)

Now the transform function is defined as݂௅ ሺܺ௅ ሻ ൌ ܺ଴ + ሺܺ஽ିଵ െ ܺ଴ ሻ݂ܿ݀௅ ሺܺ௜ ሻ
and

݂௎ ሺܺ௎ ሻ ൌ ܺ஽ + ሺܺ௅ିଵ െ ܺ஽ ሻ݂ܿ݀௎ ሺܺ௜ ሻ

(23)

(24)

And output image
ܻ ൌ {ܻሺ݅, ݆ሻ}
ൌ ݂௅ ሺܺ௅ ሻ ∪ ݂௎ ሺܺ௎ ሻ

(25)

16


݂௅ ሺܺ௅ ሻ Equalize the sub-image ܺ௅ over the range [ܺ଴ , ܺ஽ିଵ ] whereas ݂௎ ሺܺ௎ ሻ equalize the
subimage ܺ௎ over the range [ܺ஽ , ܺ௅ିଵ ].

D. Recursive Mean Separate Histogram Equalization (RMSHE):This method proposes the generalization of the Brightness preserving bi-histogram equalization
(BBHE) to overcome such limitation and provide not only better but also scalable brightness
preservation. BBHE separates the input histogram into two based on its mean before equalizing
then independently while the separation is done only once in BBHE. The RMSHE method
proposes to perform the separation recursively, separate each new histogram further based on
their respective means. Its recursive nature implies scalable preservation which is very useful in
consumer electronic products [8].

E. Minimum Mean Brightness Error Bi-Histogram Equalization in Contrast
Enhancement (MMBEBHE):This method proposes the novel extension of BBHE referred to as Minimum mean brightness
error bi-histogram equalization. It performs the separation based on the threshold level which
would return minimum absolute mean brightness error (AMBE) [9]. MMBEBHE works in these
three steps defined by the following1). First it calculates the AMBE for each of the threshold level.
2). Then it finds the minimum threshold level, XT that yield minimum MBE,
3). Finally it separates the input histogram based on the XT found in step 2 and applies histogram
equalization each of the separated histograms as in BBHE.

F. Recursive Sub-Image Histogram Equalization Applied To Gray Scale
Image (RSIHE):The recursive sub-image histogram equalization (RSIHE) is a contrast enhancement method
based on the histogram equalization. This method is generalization of DSIHE. The RSIHE
method uses cumulative probability density equal to 0.5 for separating the input histogram into
the sub-histogram [10]. This process can be applied recursively for a specified number of
recursion levels to divide the image equally into sub-images. Each sub image Xi will contain the
same number of pixels. The recursive nature of this method offers scalable brightness
preservation.

G. Recursively Separated and Weighted Histogram Equalization for
Brightness Preservation and Contrast Enhancement (RSWHE):This is a new histogram equalization method named recursively separated and weighted
histogram equalization for brightness preservation and contrast enhancement (RSWHE). To
enhance the image contrast as well as preserve the image brightness. RSWHE consists of three
modules Histogram segmentation, Histogram weighting, and Histogram equalization module
[12].

17

1.

Histogram segmentation module:This module decomposes the input image histogram in the same way as RMSHE of RSIHE
does. It divides the input histogram H(X) recursively up to the some precise recursion level.
Histogram segmentation is done on the two ways first one is based on the mean and

another based on the median of the sub-histograms.
2.

Histogram weighting module:Now we know that at recursion level r, the histogram segmentation module has generated 2r
sub histograms, i.e. ‫ܪ‬௜௥ ሺܺሻ, {1 ൑ ݅ ൑ 2௥ }. The histogram weighting module specially
modifies the probability density function of each sub histogram using the normalized power
law function.
For each sub-histogram ‫ܪ‬௜௥ ሺܺሻ, corresponding original PDF ‫݌‬ሺܺ௞ ሻ, the weighted PDF
‫݌‬௪ ሺܺ௞ ሻ, probability density equation is described in (26)௣ሺ௑ೖ ሻି௣೘೔೙ ఈ೔
ቁ
೘ೌೣ ି௣೘೔೙

‫݌‬௪ ሺܺ௞ ሻ ൌ ‫݌‬௠௔௫ ቀ ௣

+ ߚ, ሺ‫ܮ‬௜ ൑ ݇ ൑ ܷ௜ ሻ

(26)

Where,
‫݌‬௠௔௫ ܽ݊݀ ‫݌‬௠௜௡ are maximum and minimum probability value from original histogram
respectively. ߙ௜ is an accumulative probability value for ith sub-histogram ‫ܪ‬௜௥ ሺܺሻ. ߙ௜ is
calculated for each sub-histogram.
௎

೔
ߙ௜ ൌ ∑௞ୀ௅೔ ‫݌‬ሺܺ௞ ሻ

(27)

ߚ is a value which is ≥ (grater then) 0. The degree of mean brightness and contrast
enhancement of output image can be controlled by adjusting ߚ. It is experimentally found that
the output image with the satisfactory quality can be produced, when ߚ is around
‫݌‬௠௔௫ . |ܺெ − ܺீ |⁄(ܺ௠௔௫ − ܺ௠௜௡ ), where ܺ௠௔௫ ܽ݊݀ ܺ௠௜௡ are the greatest and the least gray
levels of the input imageܺ, respectively [12]. After weighting process normalization is
required to normalize the weighted histogram.
3.

Histogram equalization Module:The resultant weighted and normalized PDF called ‫݌‬௪௡ (ܺ௞ ), where the ith sub histogram is
‫ܪ‬௜௥ (ܺ) where (1 ≤ ݅ ≤ 2௥ ) over the range [ܺ௅೔ , ܺ௎೔ ]. The task of histogram equalization
module is to separately equalize each of all 2௥ sub histogram by using the histogram
equalization method. Now finally combining all the resultant sub-images produce output of
RSWHE method.

2. EXPERIMENTAL RESULTS
We have processed many images to show the performance of different histogram equalization
techniques discussed above. We have selected some standard images and apply HE, BBHE,
DSIHE, RMSHE, RSIHE, MMBEBHE, RSWHE methods to evaluate their performance. The
18


performance is assessed on the basis of absolute mean brightness error (AMBE), peak signal to
noise ratio (PSNR), and structure similarity index measure (SSIM).
•

Absolute Mean Brightness Error:Absolute mean brightness error (AMBE) is used to evaluate Brightness preservation in
processed image.
AMBE is defined as‫ܧܤܯܣ‬ሺܺ, ܻሻ ൌ |ܺெ െ ܻெ |

(28)

Where, ܺெ is mean of the input image ܺ ൌ {‫ݔ‬ሺ݅, ݆ሻ}
And ܻெ is mean of the output image ܻ = {‫.})݆ ,݅(ݕ‬
Minimum brightness error means batter brightness preservation.
•

Peak signal to noise ratio:For peak signal to noise ratio (PSNR) assume an input image is ܺ(݅, ݆) which contains ‫ܰ × ܯ‬
pixels and the processed image ܻ(݅, ݆).
First compute the Mean Squared Error (MSE),
‫= ܧܵܯ‬

∑ಾ ∑ಿ |௑(௜,௝)ି௒(௜,௝)|మ
೔సభ ೕసభ

(29)

ெ×ே

Now peak signal to noise ratio (PSNR)
ܴܲܵܰ = 10 logଵ଴
•

(௅ିଵ)మ
ெௌா

(30)

Structured Similarity index:The structural similarity index is a method for measuring the similarity between two images.
The SSIM index is a full reference metric, in other words, the measuring of image quality
based on an initial uncompressed or distortion-free image as reference. Here we define the
structural similarity index:
(ଶఓ ఓ ା஼ )(ଶఙ

ା஼ )

ೣ
ೣ೤
మ
ܵܵ‫( = )ݕ ,ݔ(ܫ‬ఓమ ାఓమ೤ భ మ ାఙమ ା஼
ା஼ )(ఙ
ೣ

೤

భ

ೣ

೤

మ)

(31)

Where ߤ௫ is the mean of image X, ߤ௬ is mean of Y, ߪ௫ is standard deviation of image x, ߪ௬ is
standard deviation of image y, ߪ௫௬ is squre root of covariance of image x and y, and C1, C2
are contents.

19


•

Entropy:The entropy is a valuable tool to measure the richness of the details in the output image. For a
given PDF p, entropy Entropy[p] is computed by (32)
‫∑ − = ]݌[ݕ݌݋ݎݐ݊ܧ‬௅ିଵ ‫ܺ(݌‬௞ ) log ଶ ‫ܺ(݌‬௞ )
௞ୀ଴

(32)

Analysis Results Tables:
Table-1: For Mean Brightness Error (AMBE)

Table-2: For Peak Signal to Noise Ratio (PSNR)

(a)

(b)

Fig.3. (a) Shows chart of AMBE for different HE methods and (b) shows chart of PSNR for different HE
methods

20

Table-3: For Structure Similarity Index (SSI)

Table-4: For Entropy

Assessment of Brightness Preservation:
The results shown in the Table-1 presents the performance of brightness preservation of various
methods discussed in this paper. Based on the observation of Table-1 we see that RSWHE-M
(Mean based) is best in brightness preservation. RSWHE-D (Median based) is second best
method for brightness preservation.
Assessment of contrast enhancement:
The results shown in the Table-2 presents the PSNR (Peak Signal to Noise Ratio) values for
various methods applied to some standard images. PSNR is a metrics for image quality
assessment. The greater the PSNR, the better the image quality is. Based on the observation of
Table-2 we see that the RSWHE-M (Mean based) has the greater PSNR values of each image and
an average PSNR for all images is also has greater value as compared to other histogram
equalization methods. The RSWHE-D also produces better results as compared to other methods.

21


Table 4, presents the Entropy values compared with original image entropy to other HE based
methods for some standard images. Entropy is used to measure the richness of details in image.
richness
RSWHE-M and RSWHE-D performs batter then others in terms of entropy.
D
Inspection of visual quality:
In addition with brightness preservation and contrast enhancement an Image quality is also an
important factor in image processing. The processed image should be visually acceptable to
human eye and should have natural appearance.
We have tested no. of images with all of methods discussed in this paper. Some of them are
mages
presented here. Figure-4 shows original image of “Couple” with processed images by HE, BBHE,
4
DSIHE, RMSHE, RSIHE, MMBEBHE, RSWHE M (Mean based) and RSWHE-D (Median
RSWHE-M
RSWHE
based) methods. Figure-5 shows the result images of “Einstein”.
5

(a) Original Image

(b) HE

(c) BBHE

(d) DSIHE

(e) RMSHE (r=2)

(f) RSIHE (r=2)

(g) MMBEBHE

(h) RSWHE-M (r=2)
M

(i) RSWHE-D (r=2)
Fig. 4. Results of all methods tested on the image of ‘Couple’.

22


(a) Original Image

(b) HE

(c) BBHE

(d) DSIHE

(e) RMSHE(r=2)

(f) RSIHE(r=2)

(g) MMBEBHE

(h) RSWHE-M (r=2)
RSWHE

(i) RSWHE-D (r=2)
Fig. 5. Results of all methods tested on the image of ‘Einstein’.

3. CONCLUSION
This paper presents comparative study of different histogram equalization based image
enhancement methods. After observation of experimental results for brightness preservation we
observed that the brightness preservation is not handled well by HE, BBHE and DSIHE, but it
can be handled properly by RMSHE, RSIHE, and RSWHE. RSWHE
RSWHE-M offers better brightness
ter
preservation, better contrast enhancement and better structure similarity index as compared to
other methods. RSWHE-D is the second best method for brightness preservation and contrast
D
enhancement. The RSWHE method also offers scalable brightness preservation because of its
brightness
recursive nature. From our study it is observed that a huge work has already been done in this
field but still there exist much space for future work.

ACKNOWLEDGMENT
We would like to thank Dr. Song Der Chen, Putra Malaysia Univ., Serdang, Malaysia for giving
help in MMBEBHE[9].

23


REFERENCES
[1]
[2]
[3]

[4]
[5]
[6]
[7]

[8]

[9]

[10]
[11]

[12]

[13]

[14]

[15]

[16]
[17]

[18]

[19]

[20]

Wahab A, Chin SH, and Tan SH, “Novel Approach To Automated Fingerprint Recognition”, IEE
Proceedings Vision, Image and Signal Processing, (1998), vol. 145, pp. 160-166.
Pizer SM, “The Medical Image Display and Analysis Group at The University of North Carolina:
Reminiscences and Philosophy”, IEEE Trans. Med. Image, (2003), vol. 22, pp. 2-10.
Torre A, Peinado AM, Segura JC, Perez-Cordoba JL, Benitez MC, and Rubio AJ, “Histogram
Equalization of Speech Representation for Robust Speech Recognition”, IEEE Trans. Speech Audio
Processing, (2005), vol. 13, pp. 355-366.
Scott E Umbaugh, Computer Vision and Image Processing, Prentice Hall: New Jersey 1998, pp. 209.
Gonzalez RC and Woods RE, “Digital Image Processing” Pearson Education Pvt. Ltd, Second
Edition, Delhi, (2003).
Kim YT, “Contrast Enhancement Using Brightness Preserving Bi-Histogram Equalization”,
Consumer Electronics, IEEE Transactions on, (1997), vol. 43, no.1, pp.1-8.
Wang Y, Chen Q, and Zhang B, “Image Enhancement Based On Equal Area Dualistic Sub-Image
Histogram Equalization Method”, Consumer Electronics, IEEE Transactions on, (1999), vol. 45, no.
1, pp. 68-75.
Chen SD and Ramli A, “Contrast Enhancement Using Recursive Mean-Separate Histogram
Equalization For Scalable Brightness Preservation”, Consumer Electronics, IEEE Transactions on,
(2003), vol. 49, no. 4. pp. 1301-1309,
Chen SD and Ramli A, “Minimum Mean Brightness Error Bi-Histogram Equalization in Contrast
Enhancement”, Consumer Electronics, IEEE Transactions on, Nov (2003), vol. 49, no. 4, pp. 13101319.
Sim KS, Tso CP, and Tan YY, “Recursive Sub-Image Histogram Equalization Applied To Gray Scale
Images”, Pattern Recognition Letters, Feb (2007), vol. 28, pp. 1209-1221.
Ibrahim H, Kong NSP, “Brightness Preserving Dynamic Histogram Equalization For Image Contrast
Enhancement”, Consumer Electronics, IEEE Transactions on, Nov (2007), vol. 53, no. 4, pp. 17521758.
Kim M, Chung MG, “Recursively Separated and Weighted Histogram Equalization for Brightness
Preservation and Contrast Enhancement”, Consumer Electronics, IEEE Transactions on, Aug (2008),
vol. 54, no. 3, pp. 1389-1397.
Wang Q and Ward RK, “Fast Image/Video Contrast Enhancement Based on Weighted Threshold
Histogram Equalization”, Consumer Electronics, IEEE Transactions on, May (2007), vol. 53, no. 2,
pp. 757-764.
Abdullah-Al-Wadud M, Kabir Md.H, Dewan MAA, and Chae O, “A Dynamic Histogram
Equalization for Image Contrast Enhancement”, IEEE Trans. Consumer Electronics, May (2007), vol.
53, no. 2, pp. 593-600.
Zimmerman J, Pizer S, Staab E, Perry E, McCartney W, and Brenton B, “Evaluation Of The
Effectiveness Of Adaptive Histogram Equalization For Contrast Enhancement”, IEEE Trans. Medical
Imaging, (1988), pp. 304-312.
Caselles V, Lisani JL, Morel JM, and Sapiro G, “Shape Preserving Local Contrast Enhancement”, in
Proc. Int. Conf. Image Processing, (1997), pp. 314-317,.
Kim JY, Kim LS, and Hwang SH, “An Advanced Contrast Enhancement Using Partially Overlapped
Sub-Block Histogram Equalization”, IEEE Transactions on Circuits and Systems for Video
Technology, (2001), vol. 11, pp. 475-484.
Reza AM, “Realization of The Contrast Limited Adaptive Histogram Equalization (Clahe) For RealTime Image Enhancement”, Journal of VLSI signal processing systems for signal, image and video
technology, (2004), vol. 38, no. 1, pp. 35-44.
Wang C and Zhongfu Ye, “Brightness Preserving Histogram Equalization With Maximum Entropy: A
Variational Perspective”, IEEE Trans. Consumer Electronics, Nov (2005) vol. 51, no. 4, pp. 13261334.
Menotti D, Najman L, Facon J, and Araújo A, “Multi Histogram Equalization Methods For Contrast
Enhancement And Brightness Preserving”, IEEE Trans. Consumer Electronics, Aug (2007), vol. 53,
no. 3, pp. 1186-1194.
24

[21] Park GH, Cho HH, and Choi MR, “A Contrast Enhancement Method Using Dynamic Range Separate
Histogram Equalization”, IEEE Trans. Consumer Electronics, Nov (2008), vol. 54, no. 4, pp. 19811987.
[22] “Image Database”, online available: http://decsai.ugr.es/cvg/index2.

AUTHORS
Omprakash Patel
He was born in Jabalpur, India. He received his Bachelor of Engineering degree in
Computer Science & Engineering from GGITS Jabalpur affiliated from RGPV Bhopal,
in 2010. He is currently pursuing Master of Technology (M.Tech.) in Computer
Science & Application from School of Information Technology, UTD, RGPV, Bhopal,
India. His research interest includes image enhancement and medical image processing.

Yogendra P.S. Maravi
He obtained his Bachelor’s degree in Information Technology from UIT, RGPV,
Bhopal. He received his M.Tech degree in Computer Science from School of
Computer Science, DAVV, Indore. He is working as Assistant Professor in School of
Information Technology, UTD, RGPV, Bhopal, India.

Dr. Sanjeev Sharma
He is working as Associate Professor in the School of Information Technology, UTD,
RGPV, Bhopal, India. His research interest is in Mobile Computing. He is a member
of Computer Society of India ACM, Computer Science Teachers Association (CSTA).

25

A comparative study of histogram equalization based image enhancement techniques for brightness preservation and contrast enhancement

More Related Content

What's hot (19)

Viewers also liked (19)

Similar to A comparative study of histogram equalization based image enhancement techniques for brightness preservation and contrast enhancement (20)

Recently uploaded (20)

A comparative study of histogram equalization based image enhancement techniques for brightness preservation and contrast enhancement