Automatic Method for Contrast Enhancement of Natural Color Images

Transcription

J Electr Eng Technol.2015; 10(?): 30-40
http://dx.doi.org/10.5370/JEET.2015.10.2.030
ISSN(Print) 1975-0102
ISSN(Online) 2093-7423
Shyam Lal**, A. V. Narasimhadhan* and Rahul Kumar†
Abstract – The contrast enhancement is great challenge in the image processing when images are
suffering from poor contrast problem. Therefore, in order to overcome this problem an automatic
method is proposed for contrast enhancement of natural color images. The proposed method consist of
two stages: in first stage lightness component in YIQ color space is normalized by sigmoid function
after the adaptive histogram equalization is applied on Y component and in second stage automatic
color contrast enhancement algorithm is applied on output of the first stage. The proposed algorithm is
tested on different NASA color images, hyperspectral color images and other types of natural color
images. The performance of proposed algorithm is evaluated and compared with the other existing
contrast enhancement algorithms in terms of colorfulness metric and color enhancement factor. The
higher values of colorfulness metric and color enhancement factor imply that the visual quality of the
enhanced image is good. Simulation results demonstrate that proposed algorithm provides higher
values of colorfulness metric and color enhancement factor as compared to other existing contrast
enhancement algorithms. The proposed algorithm also provides better visual enhancement results as
compared with the other existing contrast enhancement algorithms.
Keywords: Contrast enhancement, adaptive histogram equalization, sigmoid function, and lightness
enhancement.
1. Introduction
Common and frequently serious disagreements exist
between recorded natural color images and the direct
observation of natural scenes. The human perception is very
excels in constructing a visual representation with vibrant
color and detail across the wide range of photometric levels
due to variations in the light. In addition to the human
vision computers color are to be relatively independent of
spectral variations in illumination [1]. When we want to
display a color image on a display device, either the low
intensity as well as medium intensity areas, which are
underexposed, or the high intensity areas, which are
overexposed, cannot be seen by observer. In order to avoid
this problem, a number of color contrast enhancement
techniques have been developed during the past decades.
The color contrast enhancement techniques are commonly
used in various applications where subjective quality of
the image is very important. The objective of image
enhancement is to improve visual quality of image
depending on the application conditions. Contrast is an
important factor for any individual estimation of image
quality. It can be used as controlling tool for documenting
and presenting information collection during examination.
The contrast enhancement of image refers to the amount of
color differentiation that exists between various features
†
Corresponding Author: Department of E&C Engg., NITK Surathkal,
India (shyam.mtec@gmail.com)
*
Department of E&C Engg., NITK Surathkal, India ({dhan257,
rrkmalik}@ gmail.com)
Received: March 30, 2014; Accepted: November 16, 2014
30
in digital images. It is the range of the brightness present
in the digital image. The images having a higher contrast
level usually display a larger degree of color scale
difference as compared to lower contrast level. The contrast
enhancement is a process that allows image features to
show up more visibly by making best use of the color
presented on the display devices. During the last decade a
large number of contrast enhancement algorithms have
been developed for color contrast enhancement of images
for various applications. These are histogram equalization
[2], global histogram equalization [3], local histogram
equalization [4], adaptive histogram equalization and
Contrast Limited Adaptive histogram equalization [5, 6],
and other techniques and algorithms [7-38]. One of the most
widely used algorithms is global histogram equalization,
the basic idea of which is to adjust the intensity histogram
to approximate a uniform distribution. It treats all regions
of the image equally and, thus often yields poor local
performance in terms of detail preservation of image.
The outline of this paper is as follows. Section 2
describes literature review. Section 3 describes proposed
method for contrast enhancement of natural color images.
Section 4 gives simulation results and discussions to
demonstrate the performance of the proposed method.
Finally, conclusion is drawn in section 5.
2. Literature Review
The existing contrast enhancement techniques for
mobile communication and other real time applications is
Copyright ⓒ The Korean Institute of Electrical Engineers
This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/
licenses/by-nc/3.0/)which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Shyam Lal, A.V. Narasimhadhan and Rahul Kumar
fall under two broad categories mainly contrast shaping
based methods and histogram equalization based methods
[2]. These methods are derived from digital image
processing. These methods may lead to over-enhancement
and other artifacts such as flickering, and contouring. The
contrast shaping based methods are worked on by
calculating an input-output luminance curve defined at
every luminance level. The shape of the curve must depend
on the statistics of the image frame being processed. For
example, dark images would have a dark stretch curve
applied to them. Although contrasts shaping based methods
are the most popular methods used in the consumer
electronics industry but they cannot provide a localized
contrast enhancement which is desirable. For example,
when a dark stretch is performed, bright pixels become
brighter. However, a better way to enhance darker images
is to stretch and enhance the dark regions, while leaving
brighter pixels untouched [2][38]. A very popular
technique for contrast enhancement of image is histogram
equalization technique [2-4]. A histogram equalization is a
technique that generates gray map which change the
histogram of image and redistributing all pixel values to be
as close as possible to user specified desired histogram.
This technique is useful for processing images that have
little contrast with equal number of pixels to each the
output gray levels. The histogram equalization (HE) is a
method to obtain a unique input to output contrast transfer
function based on the histogram of the input image which
results in a contrast transfer curve that stretches the peaks
of the histogram (where more information is present) and
compresses the troughs of the histogram (where less
information is present) [2]. Therefore it is a special case of
contrast shaping technique. As a standalone technique,
histogram equalization is used extensively in medical
imaging, satellite images and other applications where the
emphasis is on pattern recognition and bringing out of
hidden details. Thus histogram equalization results in too
much enhancement and artifacts like contouring which is
unacceptable in consumer electronics [5-6].
During the last decade a number of techniques have been
proposed by various researchers to deal with these problems.
In [8], the histogram is divided into two parts based on the
input mean, and each part is equalized separately. This
preserves the mean value of image to a certain extent. In
[9], each peak of the histogram is equalized separately. An
adaptation of HE, termed as contrast limited adaptive
histogram equalization (CLAHE) [6] divides the input image
into a number of equal sized blocks and then performs
contrast limited histogram equalization on each block. The
contrast limiting is done by clipping the histogram before
histogram equalization. This tends to tone down the over
enhancement effect of histogram equalization and gives a
more localized enhancement. However it is much more
computationally intensive than histogram equalization. If
the blocks are non-overlapping, an interpolation scheme is
needed to prevent blocky artifacts in the output picture.
Therefore overlapping blocks can solve this problem (every
pixel is replaced by the histogram equalization output using
a neighborhood) but it is more computationally intensive
than using non-overlapping blocks. So the CLAHE also
requires a field store. Finally one more contrast enhancement
method that is homomorphic filter is proposed in spatial
domain [2]. In this filter images normally consist of light
reflected from objects. The basic nature of the image may
be characterized by two components: (1) the amount of
source light incident on the scene being viewed, and (2) the
amount of light reflected by the objects in the scene but
this method does not provide good image quality [5].
Another method is histogram specification which takes a
desired histogram by which the expected output image
histogram can be controlled [2]. However specifying the
output histogram is not a smooth task as it varies from
image to image. In [10] D. J. Jobson, Z. Rahman, and G.
A. Woodell introduced new algorithm to improve the
brightness, contrast and sharpness of an image. This
algorithm performs a non-linear spatial transform that
provides simultaneous dynamic range compression [11].
The performance of this method is compared with other
existing enhancement techniques such as histogram
equalization and homomorphic filtering [12]. In [13] B. V.
Funt, K. Barnard, M. Brockington, and V. Cardei have
investigated the Multi Scale Retinex algorithm approach
for image enhancement purpose, they explored the effect of
processing from theoretical standpoint [13] and in the same
year they modified the multi-scale retinex approach to
image enhancement such that the processing is more
justified from a theoretical standpoint they suggested a new
algorithm with fewer arbitrary parameters and prove it is
more flexible [14]. In [15] A. A. Bayaty has suggested a
new method to calculate image contrast and evaluate image
quality depending on computing the image contrast in edge
regions. Author has introduced robust quantitative measures
to determine image quality, and after that estimated the
efficiency of the various techniques in image processing
applications. In [16] L. Tao, M. J. Seow and V. K. Asari have
proposed image contrast enhancement method to improve
the visual quality of digital images that exhibit dark
shadows due to the limited dynamic ranges of imaging and
display devices which are incapable of handling high
dynamic range scenes. The proposed technique processes
images by applying two separate steps: dynamic range
compression and local contrast enhancement. Dynamic
range compression is a neighborhood dependent intensity
transformation which is able to enhance the luminance in
dark shadows while keeping the overall tonality consistent
with that of the input image. In [17] A. J. A. Dalawy has
studied the TV satellite images. These images were the
same with respect to the type on the three satellites. The
analyzing these images was done statistically by finding
the statistics distribution and studying the relations
between the mean and the standard deviation of the color
compound (RGB) and light component (L) for the image as
http://www.jeet.or.kr │ 31
whole and for the extracted homogeneous regions. Also
author studied the contrast of image edges depending on
sobel operator in neighbor area to the edges and studied the
contrast as function for edge finding threshold and found
that the Hotbird has the best results.
In [18] N. Hassan and N. Akamastu have proposed new
approach for contrast enhancement using sigmoid function.
The objective of this new contrast enhancer is to scale the
input image by using sigmoid function. However this
method is also have some side effects. In order to improve
the performance of above mentioned technique another
algorithm that is exact histogram specification (EHS) is
proposed for contrast enhancement of images [19].
However this method is also have some side effects. In
order to provide better result another technique that is
brightness preserving dynamic fuzzy histogram equalization (BPDFHE) is proposed [20]. This technique is the
modification of the brightness preserving dynamic
histogram equalization technique to improve its brightness
preserving and contrast enhancement abilities while
reducing its computational complexity. This technique uses
fuzzy statistics of digital images for their representation
and processing. Therefore, representation and processing of
images in the fuzzy domain enables the technique to handle
the inexactness of gray level values in a better way which
results provide improved performance. However this
technique is also having some side effects. In [21] Celik
and Jahjadi proposed contextual and variational contrast
enhancement for image. This algorithm enhances the
contrast of an input image using interpixel contextual
information. This algorithm uses a 2-D histogram of the
input image constructed using a mutual relationship
between each pixel and its neighboring pixels. A smooth 2D target histogram is obtained by minimizing the sum of
frobenius norms of the differences from the input
histogram and the uniformly distributed histogram. The
enhancement is achieved by mapping the diagonal
elements of the input histogram to the diagonal elements of
the target histogram. This algorithm produces better
enhanced images results as compared to other existing
state-of-the-art algorithms but this method is also have
some side effects.
In [22] H. Hu and G. Ni proposed an improved retinex
algorithm for image enhancement purpose. This algorithm
has provides very good performance for specific color
images in terms of color constancy, contrast enhancement
and computational cost. However, this algorithm is done in
HSV color space rather than RGB space. Therefore an
additional step is necessary to convert an image from RGB
to HSV color space and vice-versa. However this algorithm
is also have some side effects. In [23] Y. Terai et al.
proposed a retinex model for color image contrast
enhancement purpose. In this algorithm, the luminance
signal is processed in order to reduce the computation time
without changing color components from one format to
other format. But the computation time of this algorithm is
32 │ J Electr Eng Technol.2015;10(1): 30-40
still large due to large scale Gaussian filtering. The
algorithm is only suitable for gray images. However this
algorithm is also have some side effects. In [24] L. He et al.
proposed image enhancement algorithm based on retinex
theory. This algorithm replaces brightness value of each
pixel by the ratio of brightness value to the average values
of the neighboring pixels. The main drawbacks of this
algorithm are complexity, weakly illuminated, enlarged
dynamic range. In [25] X. Ding et al. proposed a new
enhancement algorithm for color image based on human
visual system based on adaptive filter. This algorithm
utilizes color space conversion to obtain a much better
visibility. This algorithm has provides better effectiveness
in reducing halo and color distortion. However, the main
drawback of this algorithm is computational slow algorithm.
On the other hand various researchers also proposed many
algorithms for contrast enhancement in DCT based
compressed domain such as alpha rooting (AR) [7], multi
contrast enhancement (MCE) [26], modified histogram
Equalization (MHE) [27], Adaptive Contrast Enhancement
Based on modified Sigmoid Function (ACEBSF) [28], Multicontrast Enhancement with Dynamic Range Compression
(MCEDRC) [29], Contrast Enhancement by Scaling
(CES)[30], RGB retinex theory [31] and other methods for
contrast enhancement [32-35]. In order to determine image
quality metric, many existing image quality assessment
algorithms use only limited image features.
During the past years various algorithms have been
developed by various researchers, each algorithm has its
own advantages and disadvantages. The main drawback of
many existing image algorithms are the visual quality of
the enhanced image is not good. In order to overcome this
drawback of many existing contrast enhancement algorithms,
a new method is proposed for contrast enhancement of
different types of the natural color images. The proposed
algorithm provide better contrast enhancement results as
compared to other many existing contrast enhancement
algorithms for different types of the natural color images
such as NASA images, Hperspectral images and other
types of images.
3. Proposed Algorithm
The proposed algorithm consists of two stages: in first
stage lightness component in YIQ color space is
transformed using sigmoid function, after the adaptive
histogram equalization (AHE) method [6] is applied on Y
component and in the second stage automatic color
enhancement algorithm is applied. The proposed algorithm
is abbreviated as Automatic Color Contrast Enhancement
(ACCE) Algorithm. The model of proposed method is
shown in Fig. 1.
Stage-I: In the first stage the input color image is
converted to YIQ Color space by RGB to YIQ color space
Color Space
Transformation
(RGB to YIQ)
Input Color Image
(RGB)
Consider I ( x, y ) = Y = 0.299 R + 0.587G + 0.114 B,
the normalized Intensity is given by
I n ( x, y ) =
Chromatic Components
[I], [Q]
Illumenance Component
[Y]
I ( x, y )
255
then
(2)
After this the normalized lightness value is transformed
by using sigmoid function is given by [35]
Normalized by Sigmoid
function and Applied AHE
Yp
S n ( x, y ) =
Automatic Color
Contrast Enhancement
(RGB)
Fig. 1. Block diagram of proposed algorithm
transformation because it is used in the NSTC and PAL
televisions of different countries. First advantages of this
format is that grayscale information is separated from color
data, so the same signal can be used for both color and
black and white sets. In the NTSC format, image data
consists of three components: luminance (Y), hue (I), and
saturation (Q).The first component, luminance, represents
grayscale information, while the last two components make
up chrominance (color information). Second advantage is
that it takes advantage of human color-response
characteristics. The eye is more sensitive to changes in the
orange-blue (I) range than in the purple-green range (Q),
therefore less bandwidth is required for Q than for I. The
color space transformation is given as:
I = 0.596 R − 0.275G − 0.321B
Q = 0.212 R − 0.523G − 0.311B
⎫
⎪
⎬
⎪
⎭
⎛
⎞
⎜1 + (1 − I n ( x, y ) ⎟
⎜
I n ( x, y ) ⎟⎠
⎝
(3)
The plot between Sn and In is shown in Fig. 2.
The processed lightness component is obtained by
applying adaptive histogram equalization (AHE) [6] on the
lightness component and the processed lightness
component is denoted as YP for further reference. The YIQ
to RGB color space transformation of the first stage is
defined as follows
Output Color image
Y = 0.299 R + 0.587G + 0.114 B
1
⎫
⎪
⎪
G ' = Y p − 0.2721I − 0.6474Q ⎬
⎪
B ' = Y p − 1.1070 I + 1.7046Q ⎪
⎭
R ' = Y p + 0.9563I + 0.6210Q
(4)
Stage - II: The automatic color contrast enhancement
algorithm is given bellow:
Step(1) Convert R 'G ' B ' to Y ' I 'Q '
Step(2) Set the following parameters
my _ Limit = 0.5, lower _ Limit = 0.008
upper _ Limit = 0.992; my _ Limit 2 = 0.04;
(1)
my _ Limit 3 = −0.04
Step(3) Calculate Y _ Adjust , I _ Adjust , Q _ Adjust and
find Yn, In and Qn components
After the color space transformation, the luminance
component (Y) is normalized as follows:
1
0.9
0.8
0.7
Sn
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
In
0.6
0.7
0.8
0.9
1
Fig. 2. Relationship between input lightness In versus output lightness Sn
'
Y _ Adjust = my _ Limit − Ymean
Yn = Y ' + Y _ Adjust * (1 − Y ' )
'
I _ Adjust = my _ Limit 2 − I mean
I n = I ' + I _ Adjust * (0.596 − I ' )
'
Q _ Adjust = my _ Limit 3 − Qmean
Qn = Q ' + Q _ Adjust * (0.523 − Q ' )
Step(4) Convert image YnInQn to RnGnBn (F1), the inverse
color space transformation from YnInQn to RnGnBn
color space is given as
Rn = Yn + 0.9563 I n + 0.6210Qn
Gn = Yn − 0.2721I n − 0.6474Qn
Bn = Yn − 1.1070 I n + 1.7046Qn
⎫
⎪
⎬
⎪
⎭
F = F 1*255
Step(6) Final output image can be calculated as
Fout =
Vmax − Vmin
The test natural color images used for the experiment
are available on the websites http://dragon.larc.nasa.
govt/retinex/pao/news and http://vision.seas.harvard.
edu/hyperspec/explorei.html
(5)
Step(5) Normalized image intensity which is given below
(F − Vmin )
metrics are used to compare the performance of proposed
ACCE algorithm and other existing contrast enhancement
techniques such adaptive histogram equalization (AHE)
[6], alpha rooting (AR) [7], multi contrast enhancement
(MCE) [26], modified histogram Equalization (MHE)
[27], Adaptive Contrast Enhancement Based on modified
Sigmoid Function (ACEBSF) [28], Multi-contrast Enhancement with Dynamic Range Compression (MCEDRC) [29],
Contrast Enhancement by Scaling (CES)[30], RGB retinex
theory [31].
(6)
Where Vmin= max {minimum value of RnGnBn with impact
of lower limit (0.008) }
Vmax = max {maximum value of RnGnBn with impact of
upper limit (0.992)}
The colorfulness metric (CM) is no-reference image
quality metric CM). It is suggested by Susstrunk and
Winkler [38]. The definition for this metric in the color
space is as given below. Let the red, green and blue
components of an image be denoted by R, G and B,
respectively [30]. Consider α=R-G and β=(R+G)/2-B, then
the colorfulness of the image is defined as
CM ( I ) =
In order to demonstrate the performance of proposed
ACCE algorithm, it is tested on different natural color
images such NASA color images, hyperspectral color
images and other types of natural color images. The
proposed ACCE algorithm and other existing algorithms
are implemented using MATLAB software (MATLAB 7.6,
release 2008a), and 4GB RAM with I3 Processor. Two
experiments are conducted on different natural color
images: in the first experiment the image quality metrics is
evaluated and in the second experiment visual enhancement quality of image is obtained. In order to judge the
performance of proposed ACCE algorithm the color image
quality parameters such as colorfulness metric (CM), color
enhancement factor (CEF) and CPU time are the automatic
choice for the researchers. The higher values of CM and
CEF imply that the visual quality of the enhanced image is
good and a lower value of CPU time is good. The
colorfulness metric (CM) and color enhancement factor
(CEF) are defined in Eq. (7) and Eq. (8) respectively for
natural color images. These non-reference image quality
2
α
+ σ β2 ) + 0.3*
(μ
2
α
+ μ β2 )
(7)
Where σα and σβ are standard deviations of α and β
respectively. Similarly, μα and μβ are their means.
The color enhancement factor (CEF) between output
image and input image is defined as:
CEF =
4. Simulation Results and Discussions
(σ
colorfu ln ess(CM ) of output image
colorfu ln ess(CM ) of input image
(8)
4.1 Experiment 1
In this experiment the performance of proposed ACCE
algorithm is tested on different natural color images such
NASA color images (image11.jpg, image22.jpg, image99.
jpg, and image110.jpg), Hyperspectral color images
(image33.jpg, image44.jpg, and image55.jpg) and other
natural color images (image66.jpg). The performance of
proposed ACCE algorithm has been evaluated and compared
with many existing contrast enhancement techniques such
adaptive histogram equalization (AHE) [6], alpha rooting
(AR) [7], multi contrast enhancement (MCE) [26], modified
histogram Equalization (MHE) [27], Adaptive Contrast
Enhancement Based on modified Sigmoid Function
(ACEBSF) [28], Multi-contrast Enhancement with
Dynamic Range Compression (MCEDRC) [29], Contrast
Enhancement by Scaling (CES)[30], RGB retinex theory
[31] for hyperspectral color images and other natural color
images and for NASA color images. The performance of
proposed ACCE algorithm and many other existing
contrast enhancement techniques have been evaluated and
Table 1. Comparative performance of different methods for Hyperspectral images & other natural image
Method
AHE[6]
RRT [31]
MCE DRC
[29]
CM(Original)
CM(Output)
CEF
CPU Time (second)
26.23
49.35
1.88
0.20
26.23
26.77
1.02
28.56
26.23
26.22
1.00
4.37
CM(Original)
CM(Output)
CEF
CPU Time (second)
33.49
42.76
1.28
0.22
33.49
40.63
1.21
27.94
33.49
33.56
1.00
4.45
CM(Original)
CM(Output)
CEF
CPU Time (second)
34.24
33.39
0.98
0.20
34.24
48.57
1.42
28.08
34.24
34.24
1.00
4.24
CM(Original)
CM(Output)
CEF
CPU Time (second)
33.79
33.77
1.00
0.22
33.79
29.09
0.86
40.86
33.79
34.02
1.01
5.38
Parameters
MCE [26]
image33.jpg
26.23
26.23
1.00
0.03
image44.jpg
33.49
33.49
1.00
0.02
image55.jpg
34.24
34.24
1.00
0.02
image66.jpg
33.79
33.79
1.00
0.02
AR [7]
MHE [27]
ACEBSF
[28]
CES [30]
ACCE
(Proposed)
26.23
26.57
1.01
0.08
26.23
0.72
0.03
0.14
26.23
0.45
0.02
0.36
26.23
31.87
1.21
0.06
26.23
62.16
2.37
0.34
33.49
33.35
1.00
0.09
33.49
0.73
0.20
0.16
33.49
0.68
0.02
0.34
33.49
38.21
1.14
0.08
33.49
73.39
2.19
0.36
34.24
34.23
1.00
0.08
34.24
0.73
0.20
0.14
34.24
0.22
0.01
0.34
34.24
43.18
1.26
0.06
34.24
82.33
2.40
0.34
33.79
34.23
1.01
0.09
33.79
0.72
0.20
0.16
33.79
0.43
0.01
0.39
33.79
45.07
1.33
0.08
33.79
58.19
1.72
0.41
AR [7]
MHE [27]
ACEBSF
[28]
CES [30]
ACCE
(Proposed)
51.06
50.71
0.99
0.05
51.06
0.72
0.01
0.11
51.06
0.45
0.01
0.24
51.06
53.93
1.06
0.06
51.06
56.27
1.10
0.25
22.70
22.22
0.98
0.05
22.70
0.73
0.03
0.11
22.70
0.38
0.02
0.25
22.70
31.27
1.38
0.06
22.70
31.82
1.40
0.38
16.58
16.54
1.00
0.05
16.58
0.73
0.04
0.09
16.58
0.31
0.02
0.25
16.58
26.34
1.59
0.03
16.58
27.40
1.65
0.25
16.87
16.87
1.00
0.08
16.87
0.73
0.04
0.13
16.87
0.22
0.01
0.28
16.87
21.84
1.29
0.03
16.87
30.65
1.82
0.31
17.57
17.34
0.99
0.05
17.57
0.73
0.04
0.11
17.57
0.36
0.02
0.25
17.57
30.46
1.73
0.02
17.57
25.60
1.46
0.28
Table 2. Comparative performance of different methods for NASA images
Method
AHE[6]
RRT [31]
MCE DRC
[29]
CM(Original)
CM(Output)
CEF
CPU Time (second)
51.06
51.75
1.01
0.19
51.06
50.75
0.99
8.43
51.06
50.26
0.98
5.80
CM(Original)
CM(Output)
CEF
CPU Time (second)
22.70
26.16
1.15
0.19
22.70
23.49
1.03
8.63
22.70
22.34
0.98
2.53
CM(Original)
CM(Output)
CEF
CPU Time (second)
16.58
27.05
1.63
0.17
16.58
22.71
1.37
9.30
16.58
16.71
2.43
1.01
CM(Original)
CM(Output)
CEF
CPU Time (second)
16.87
22.56
1.34
0.20
16.87
19.42
1.15
14.16
16.87
17.05
1.01
3.48
CM(Original)
CM(Output)
CEF
CPU Time (second)
17.57
23.22
1.32
0.19
17.57
21.62
1.23
8.85
17.57
17.65
1.01
2.46
Parameters
MCE [26]
image33.jpg
51.06
51.06
1.00
0.01
image22.jpg
22.70
22.70
1.00
0.02
image88.jpg
16.58
16.58
1.00
0.01
image99.jpg
16.87
16.87
1.00
0.01
image100.jpg
17.57
17.57
1.00
0.06
compared in terms color fullness metric (CM), color
enhancement factor (CEF) and CPU time which are given
in Table 1 and Table 2. From Table 1 and Table 2, it is
observed that the proposed ACCE algorithm provides
higher values of CM) and (CEF) as compared to other
many existing contrast enhancement techniques. And also
proposed ACCE algorithm provides lower CPU time and
litter more CPU time as compared to other existing
algorithm as given in Table 1 and Table 2.
4.2 Experiment 2
This experiment visualizes subjective image enhancement performance. The enhanced contrast of NASA images,
Hyperspectral images and other natural color images have
(A).Original image
(B).Output result RRT
(C).Output result of
AHE
(D).Output result of Alpha
Rooting
(E).Output result of MCE
(F).Output result of
MCEDRC
(G).Output result of
MHE
(H).Output result of
ACEBSF
(I).Output result of CES
(J).Output result of
ACCE algorithm
Fig. 3. Visual Enhancement results of different algorithms for image11.jpg
(A).Original image
AHE
Rooting
MCEDRC
MHE
ACEBSF
ACCE algorithm
(A).Original image
AHE
Rooting
MCEDRC
MHE
ACEBSF
ACCE algorithm
(A).Original image
AHE
Rooting
MCEDRC
MHE
ACEBSF
ACCE algorithm
(A).Original image
AHE
Rooting
MCEDRC
MHE
ACEBSF
ACCE algorithm
(A).Original image
AHE
Rooting
MCEDRC
MHE
ACEBSF
ACCE algorithm
been compared with result of proposed ACCE algorithm
and many other existing contrast enhancement techniques.
The visual contrast enhancement results of proposed ACCE
algorithm and many existing contrast enhancement
techniques have been given from Fig.3 to Fig.8. Therefore,
it can be noticed from Fig. 3(B) to Fig. 3(J), Fig. 4(B) to
Fig. 4(J), Fig. 5(B) to Fig. 5(J), Fig. 6(B) to Fig. 6(J) Fig.
7(B) to Fig. 7(J) and Fig. 8(B) to Fig. 8(J) that proposed
ACCE algorithm gives better color contrast enhancement
results as compared to other existing contrast enhancement
techniques.
5. Conclusion
In this paper, an automatic color contrast enhancement
algorithm has been proposed for image enhancement
purpose for various applications. The proposed method has
been tested on different types of natural color images such
as NASA images, Hyperspectral images and other types of
natural images. The subjective enhancement performance
of proposed ACCE algorithm has been evaluated and
compared with other state-of-art contrast enhancement
techniques for different natural color images in terms of
colorfulness metric (CM) and Color enhancement factor
(CEF). The simulation results demonstrated that the
proposed ACCE algorithm provides better color
enhancement quality parameters such as Colorfulness
metric (CM) and Color enhancement factor (CEF) and also
provided better visual enhancement results as compared to
other state-of-art contrast enhancement techniques for
different natural color images. Therefore, the proposed
ACCE algorithm performs very effectively for the contrast
enhancement of different natural color images. The
proposed ACCE algorithm can also be used for many other
images such as remote sensing images and even real life
photographic pictures suffer from poor contrast as well as
medium contrast problems during its acquisition.
References
[1] D. Jabson, Z. Rahman and G. A. Woodel, “A MultiScale Retinex for Bridging the Gap Between Color
Images and the Human Observation of Scenes,” IEEE
Trans. Image Processing, vol. 6, pp. 965-976, July
1997.
[2] R.C. Gonzalez and R. E Woods, “Digital Image
Processing,” Addison-Wesley Publishing Company,
2nd edition, 1992.
[3] J. A Stark, “Adaptive Contrast Enhancement using
Generalization of Histogram Equalization,” IEEE
Trans. on Image Processing, vol. 9, no. 5, pp. 889-906,
2000.
[4] V. Caselles, J. L. Lisani, J. M. Morel and G. Sapiro,
“Shape Preserving Local Histogram Modification,”
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
IEEE Trans. on Image Processing, vol. 8, no. 2, pp.
220-230, 1998.
S. M. Pizer, E. P. Amburn, J. D. Austin, R. Cromartie,
A. Geselowitz, T. Greer, B. T. H Romeny, J. B
Zimmerman and K. Zuiderveld, “Adaptive Histogram
Equalization and its Variations,” Computer Vision,
Graphics and Image Processing, vol. 39, no. 3, pp.
355-368, 1987.
K. Zuiderveld, “Contrast Limited Adaptive Histogram
Equalization,” Chapter VIII.5, Graphics Gems IV,
Cambridge, MA, Academic Press, pp. 474-485, 1994.
S. Aghagolzadeh and O. K. Ersoy, “Transform Image
Enhancement,” Optical Engineering, vol. 31, pp. 614626, 1992.
Y. T Kim, “Contrast Enhancement using Brightness
Preserving Bi-histogram Equalization,” IEEE Trans.
on Consumer Electronics, vol. 43, no. 1, pp. 1-8, 1997.
S. D. Chen and A. R. Ramli, “Preserving Brightness
in Histogram Equalization Based Contrast Enhancement Techniques,” Digital Signal Processing, vol. 14,
no. 5, pp. 413-428, 2004.
Z. Rahman, G. A. Woodell, and D. J. Jobson, “A
Comparison of the Multiscale Retinex with Other
Image Enhancement Techniques,” In Proceedings of
the IS &T 50th Anniversary Conference, May 1997.
Z. Rahman, G. A. Woodell and D. J. Jobson, “A
Comparison of the Multiscale Retinex with Other
Image Enhancement Techniques,” Proceedings of the
IS & T 50th Anniversary Conference, 1997.
D. J. Jobson, Z. Rahman and G. A. Woodell, “Properties and Performance of a Center / Surround
Retinex,” IEEE Trans. on Image Processing, vol. 6, pp.
451-462, 1996.
B. V. Funt, K. Barnard, M. Brockington and V. Cardei,
“Luminance-Based Multi-scale Retinex”, In Proceedings of 9th Congress of the International Colour
Association (AIC Color 97), Kyoto, Japan, 1997.
K. Barnard and B. Funt, “Analysis and Improvement
of Multi-Scale Retinex,” In Proceedings of the IS
&T/SID, Fifth Color Imaging Conference: Color
Science, Systems and Applications, Scottsdale,
Arizona, pp. 221-226, November 17-20, 1997.
A. Bayaty, “Adaptive Techniques for Image Con-trast
Estimation Based on Edge Detection,” Master Degree
Thesis, Physics Dep., Al-Mustansiriya Univ., 2005.
L. Tao, M. J. Seow and V. K. Asari, “Nonlinear Image
Enhancement to Improve Face,” International Journal
of Computational Intelligence Research, vol. 2, no. 4,
pp. 327-336, 2006.
J. A. Dalawy, “A Study of TV Images Quality for
Channels Broadcast Television Satellite,” Master of
Science in Physics, Physics Department, AlMustansiriya University, 2008.
N. Hassan and N. Akamatsu, “A New Approach For
Contrast Enhancement Using Sigmoid Function,” The
International Arab Journal of Information Tech-nology,
vol. 1, no. 2, pp. 221-225, 2004.
D. Coltuc, P. Bolon and J. M. Chasserym, “Exact
Histogram Specification,” IEEE Trans. on Image
Processing, vol. 15, no. 5, pp. 1143-1151, 2006.
[20] D. Sheet, H. Garud, A. Suveer, A. M. Mahadevappa
and J. Chatterjee, “Brightness Preserving Dynamic
Fuzzy Histogram Equalization,” IEEE Trans. on
Consumer Electronics, vol. 56, no. 4, pp. 2475-2480,
2010.
[21] T. Celik and T. Tjahjad, “Contextual and Variational
Contrast Enhancement,” IEEE Trans. on Image
Processing, vol. 20, no. 12, pp. 3431-3441, 2011.
[22] H. Hu and G. Ni, “The Improved Algorithm for the
Defect of the Retinex Image Enhancement”, In
Proceedings of International Conference on AntiCounterfeiting Security and Identification in Communication, pp. 257-260, July, 2010.
[23] Y. Terai, T. Goto, S. Hirano and M. Sakurai, “Color
Image Contrast Enhancement by Retinex Model,” In
Proceedings of IEEE 13th International Symposium
on Consumer Electronics, pp. 392-393, May, 2009.
[24] L. He, L. Luo, and J. Shang, “An Image Enhancement Algorithm based on Retinex Theory,” Proceedings of First International Workshop on Education
Technology and Computer Science, pp. 350-352,
March, 2009.
[25] X. Ding, X. Wang and Q. Xiao, “Color Image
Enhancement with a Human Visual System based
Adaptive Filter,” In Proceedings of International
Conference on Image Analysis and Signal Processing,
April, 2010.
[26] J. Tang, E. Peli and S. Acton, “Image Enhancement
using A Contrast Measure in the Compressed Domain,”
IEEE Signal Processing Letter, vol. 10, no. 10, pp.
289-292, 2003.
[27] H. H. Kareem, “Color image with Dim regions
Enhancement Using Modified Histogram Equalization Algorithm,” Journal of Al-Nahrain University,
vol. 15, no. 3, pp. 101-111, 2012.
[28] S. Lal, and M. Chandra, “Efficient Algorithm for
Contrast Enhancement of Natural Images,” The
International Arab Journal of Information Technology,
vol. 11, no. 1, pp. 96-103, 2014.
[29] S. Lee, “An Efficient Content Based Image Enhancement in the Compressed Domain Using Retinex
Theory,” IEEE Trans. Circuits Syst. Video Technol.,
vol. 17, no. 2, pp. 199-213, February 2007.
[30] J. Mukherjee and S. K Mitra, “Enhancement of Color
Images by Scaling the DCT Coefficients,” IEEE Trans.
on Image Processing, vol. 17, no. 10, October 2008.
[31] R. Kimmel, M. Elad, D. Shaked, R. Keshet, and I.
Sobel, “A Variational Framework for Retinex,” In
Proceedings of SPIE Electronic Imaging, vol. 4672,
2002.
[32] M. C. Hanumantharaju, M. Ravishankar, D. R.
Rameshbabu and S. Ramachandran,” Color Image
Enhancement Using Multiscale Retinex with
Modified Color Restoration Technique,” In Proceedings of International Conference on Emerging
Applications of Information Technology (EAIT-
2011) , pp. 93-97, 2011.
[33] Loza, D. Bull and A. Achim, “Automatic Contrast
Enhancement of Low-Light Images Based on Local
Statistics of Wavelet Coe cients,” In Proceedings of
17th IEEE International Conference on Image Processing, September 26-29, 2010.
[34] S. C. Huang, F. C. Cheng, and Y. S. Chiu,” Efficient
Contrast Enhancement Using Adaptive Gamma Correction With Weighting Distribution” IEEE Trans. on
Image Processing, vol. 22, no. 3, pp.1032-104, March
2013.
[35] E. Provenzi and V. Caselles, “A Wavelet Perspective
on Variational Perceptually-Inspired Color Enhancement”, International Journal of Computer Vision, vol.
106, no.2, pp. 153-171, 2014.
[36] Menon, K. Mehrotra, C. Mohan and S. Ranka,
“Characterization of a class of Sigmoid Functions
with applications to Neural Networks, Neural Networks, vol. 9, no. 5, pp. 819-8351, 1996.
[37] S. J. Sangwine and R. E. N. Horne, “The Color Image
Processing, Hand Book,” international Thomson,
1998.
[38] S. Susstrunk and S. Winkler, “Color Image Quality on
the Internet,” In Proceedings of IS & T/SPIE Electronic Imaging: Internet Imaging V, vol. 5304, pp.
118-131, 2004.
Shyam Lal is working as Assistant
Professor in the department of Electronics & Communication Engineering,
National Institute of Technology
Karnataka, Surathkal, Mangalore (KA),
India. He has more than 12 years of
teaching and research experience. He
has published more than 45 papers in
the area of Digital Image Processing and Wireless Communication & Computing at International/National Journals
& Conferences. He is Member of IEEE and other reputed
professional societies. He has been Guest Editor of International Journal of Signal & Imaging System Engineering
(IJSISE), Inderscience Publishers. His area of interest
includes Digital Image Processing, Digital Signal Processing and Wireless Communication
A.V. Narasimhadhan is working as
Assistant Professor in the department
of Electronics & Communication
Engineering, National Institute of
Technology Karnataka, Surathkal,
Mangalore (KA), India. He has more
than 2 years of teaching and research
experience. He has published more
than 12 papers in the area of Medical Imaging at
International/National Journals & Conferences. He is
Member of IEEE and other reputed professional societies.
His area of interest includes Medical Imaging and Image
Processing.
Rahul Kumar is currently pursuing
M.Tech. in Signal & Image Processing
from, National Institute of Technology,
Rourkela, Odisha, India. He has more
than 3 years of Teaching & Research
experience. His area of interest
includes Digital Image Processing,
Digital Signal Processing and Robotics.

Automatic Method for Contrast Enhancement of Natural Color Images

Transcription

Similar documents

Lecture 1

Colour Science Technical description of the i e image enhancement

Bistango Event - Diane Harkey for Board of Equalization

dinosaurs offer - figury

AMANAH IKHTIAR MALAYSIA

Staff Development at Middlesex July 2014

Digital image enhancement for recording rupestrian engravings

realce de contraste em imagens digitais usando

Here - reephamhigh.com

Decoding Innovation + Design

Leadership and Team Building - Regional Training Institute,Kolkata

Stems 43.2