International Journal of Engineering and Technical Research (IJETR)

ISSN: 2321-0869, Volume-3, Issue-6, June 2015

Low Contrast Image Enhancement using Wavelet

Transform based Algorithms: A Literature Review
Namita Naik, Dr. Agya Mishra
composition of the image. The disadvantages are, it cannot
Abstract Image Enhancement deals with enhancing the simultaneously enhance all parts of image in good manner and
image so that it provides more details. Since image clarity is very it is also difficult to automate the image enhancement
easily affected by lightening, weather or equipment that has procedure [11].
been used to capture the image. These conditions may lead to
image suffer from poor contrast and loss of information. The
main purpose of enhancement is to bring out detail that is
1.1IMAGE ENHANCEMENT TECHNIQUES:-
hidden in image or to increase contrast in a low contrast image. Histogram Equalization (HE):-
This paper presents a literature review ofsome Image
Enhancement algorithms in combination with wavelet The histogram of an image is a plot of number of occurrences
transform such as DWT(discrete wavelet transform), of gray levels in the image against the gray level values. The
SWT(singular wavelet transform), UWT(undecimated wavelet histogram provides a convenient summary of the intensities in
transform).This paper concludes that the algorithm using UWT an image. Equalisation is the process that attempts to spread
gives better result and may be the field of research. gray level in an image so that they are evenly distributed
across their range. Histogram equalization reassigns the
Index Terms AHE (adaptive histogram equalization), Image
brightness values of pixels based on image enhancement.
Enhancement, DCT, DSR (dynamic stochastic resonance),
DWT, SVD (Singular value decomposition, SWT,UWT
Histogram equalization provides more visually pleasing
results [2].
Adaptive Histogram Equalization (AHE) is used for
I. INTRODUCTION improving contrast in images. It differs from Histogram
Equalization by adaptive method that computes several
Image enhancement is to improve the interpretability or histograms and eachhistogram corresponding to a distinct
perception of information in images to provide better input for section of an image. The contrast of region for an image will
other automated image processing steps. As a result, many not be sufficiently enhanced by Histogram Equalization. AHE
techniques have developed known as image Enhancement improves this enhancement by transforming each
techniques to recover the information in an image. . The pixelwithatransformation function derived from a
enhancement methods can be broadly categorized into neighborhood region. It is used to overcome some limitations
following two methods: of global linear min- max windowing method. Thus it reduces
1. Spatial Domain Method the amount of noise in regions of the image. And also AHE
2. Frequency Domain Method have the ability forimproving the contrast of grayscale and
The spatial domain techniques, directly operates on pixels color image [11].
of an image. The pixel values are manipulated to achieve Dynamic Stochastic Resonance (DSR) [11] an external
desired enhancement [9]. The gain of spatial based domain noise of an image is considered for an image. And the
technique is that they conceptually simple to understand and Adaptive DWT based Dynamic Stochastic Resonance uses
the complexity of these techniques are low . But these internal noise for improving performance of an input image. It
techniques have difficulty in providing sufficient robustness produces output without artifacts, ringing, blocking of the
and imperceptibility requirements. In frequency domain image. The adding of noise to the input image is useful for
methods, the image is transferred into frequency domain. It non-linear systems using this technique. By using lower noise
means that, the Fourier transform of the image is computed intensities in SR mechanism the signal cannot be able to reach
first. The result of Fourier transform is multiplied with a filter the threshold value. In this technique the noise allows the
transfer function. And then the inverse Fourier transform is signal to reach the threshold value. Thus Adaptive DWT
performed to get the resultant image. Frequency domain based Dynamic Stochastic Resonance is suitable for
image enhancement is used to describe the analysis of enhance both the gray scale and coloured image [11].
mathematical functions and signals with respect to frequency Singular value decomposition(SVD) is based on a theorem
and operate directly on the transform coefficients of the from linear algebra which says that a rectangular matrix A,
image, such as Fourier transform, which is a product of three matrices that is (i) an orthogonal
discrete wavelet transform (DWT), and discrete cosine matrix UA, (ii) a diagonal matrix A and (iii) the transpose of
transform (DCT). The advantages of frequency domain are, an orthogonal matrix VA. The singular-value-based image
less computational complexity, manipulating the frequency equalization (SVE) technique is based on equalizing the
singular value matrix obtained by singular value
decomposition (SVD). SVD of an image, can be interpreted
Namita Naik, M.E Scholar, Dept. of Electronics and
Telecommunication Engineering, Jabalpur Engineering College, as a matrix, is written as follows:
Jabalpur,Madhya Pradesh, India,.
Dr.Agya Mishra, Assistant Professor, Dept. of Electronics and
Telecommunication Engineering, Jabalpur Engineering College, Jabalpur, Basic enhancement occurs due to scaling of singular values of
Madhya Pradesh, India. the DCT coefficients. The singular value matrix represents

123 www.erpublication.org
 Low Contrast Image Enhancement using Wavelet Transform based Algorithms: A Literature Review

the intensity information of input image and any change on the II. PERFORMANCE ANALYSIS
singular values change the intensity of the input image. The
main advantage of using SVD for image equalization, A
2.1 Algorithm using Directional WT and DCT:
contains the intensity information of the image[2].
In [1] DWT is a linear transformation that operates on a data
1.2 TRANSFORMS FOR IMAGE ENHANCEMENT
vector whose length is an integer power of two, transforming
Wavelet Transform (WT)
it into a numerically different vector of the same length and
DCT exploits interpixel redundancies to render excellent
Wavelet transform is capable of providing the time and
decorrelation for most natural images .
frequency information simultaneously. The frequency and
2.2 Algorithm using DWT, DCT & SVD:
time information of a signal .But sometimes we cannot know
what spectral component exists at any given time instant. The
In [2] the algorithm used gives better contrast as well as high
best we can do is to investigate that what spectral components
image quality. The effectiveness of the proposed algorithm is
exist at any given interval of time.
shown in fig.1.
Higher frequencies are better resolved in time, and lower
frequencies are better resolved in frequency. This means that,
a certain high frequency component can be located better in
time (with less relative error) than a low frequency component
& low frequency component can be located better in
frequency compared to high frequency component [8].
Discrete Wavelet Transform (DWT)
The discrete wavelet transform (DWT) is an implementation
of the wavelet transform using a discrete set of the wavelet
scales for numerical analysis and functional analysis. A
time-scale representation of a digital signal is obtained using
digital filtering techniques. In the discrete wavelet transform,
filters of different cut-off frequencies are used to analyse the
signal at different scales. If the wavelets are discretely
sampled, the resultant coefficients are called as discrete
wavelet transforms (DWT) [8].
The decomposition of images into various frequency ranges
permits the isolation of the frequency into certain sub-bands.
This process results in isolating small changes in an image
mainly in low frequency sub-band images. The 2D wavelet
decomposition of an image is performed by applying 1D (a)
DWT along the rows of the image first, and, then, the results
are decomposed along the columns. This Decomposition
results in four decomposed sub-band images referred to as
low-low (LL), low-high (LH), high-low (HL), and high-high
(HH) [2].
Stationary wavelet transforms (SWT)
SWT is also being used for the image resolution enhancement
[3]. SWT is similar DWT but does not use down sampling,
hence all the sub-bands will have the same size of the input
image[5]. The SWT is an inherently redundant scheme as the
output of each level of SWT contains the same number of
samples as the input so for a decomposition of N levels there
is a redundancy of N in the wavelet coefficients[3].
Discrete cosine transforms (DCT)
DCT converts a signal from spatial domain into a frequency
domain. DCT is real-valued and provides a better
approximation of a signal with few coefficients. This
approach reduces the size of the normal equations by
discarding higher frequency DCT coefficients. Important
structural information is present in the low frequency DCT
coefficients. Hence, separating the high-frequency DCT
coefficient and applying the illumination enhancement in the
lowfrequency DCT coefficient, it will collect and cover the
edge information from satellite images. The enhanced image
is reconstructed by using inverse DCT and it will be sharper
with good contrast [2]. (b)
Fig.1 (a) Input satellite image (b) Output satellite image [2]

124 www.erpublication.org
 International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869, Volume-3, Issue-6, June 2015
Table.1 shows the comparison between proposed
methodology and already existing techniques. In this the
Mean for satellite image has decreased but Standard
Deviation has increased.
Table.1 Comparison results between proposed methodology and already
existing technique [2]
Mean Standard
Deviation

Input satellite 179.0545 12.3043

image

DWT+SVD 157.0131 11.1123

Proposed 168.5849 29.1964 Fig. 2 VCM result [4]

technique
Table.2 VCM Results [4]
Input CT Scan 88.2227 88.1706 Image Original Enhanced

Image KP01.BMP 37 92
KP03.BMP 27 51
DWT+SVD 90.5319 90.4389
KP05.BMP 83 92

Proposed 94.3802 98.2193 KP06.BMP 37 53

technique KP07.BMP 50 64

2.5 Algorithm using Dynamic Stochastic Resonance with

2.3 Algorithm using DWT, SWT & SVD: DWT:
In [6] a dynamic stochastic resonance (DSR) based technique
In [3] satisfactorily contrast enhancement & super resolution in DWT domain is used for enhancement of very dark
images are obtained from the low resolution images but it greyscale & coloured image. This DSR based technique used
losses some information during interpolation. This problem of inherent noise of a dark image for implementing DSR. Table
losing information was overcome in [5]. In [5] along with 3.shows Distribution separation measure (DSM) values
DWT &SWT, SVD is used which improves the illumination attained by DWT based DSR technique and other various
of the image and also protects the edge information. Table contrast enhancement techniques.
2compares the PSNR values for various techniques for Table.3 DSM (and iteration values, n for DSR-based) for DWT- DSR and
resolution & contrast enhancement . The PSNR value of the other existing contrast enhancement techniques [6]
proposed technique was found to be 26.01dB for Baboon Test Test Test Test Test
Technique Test2
image. Metric 1 3 4 5 6

Table. 2-Comaparision of PSNR (dB) values for various techniques for DSM 56.97 5.5896 144.2 51.52 11.02 84.08
resolution and contrast enhancement from 128x128 to 512x512 DWT-
DSR
n 50 250 450 400 350 100
Techniques/images Lena Baboon Peppers
DSM 32.03 4.5758 92.50 8.012 4.463 88.40
Bilinear 24.83 20.61 25.66 DFT-
DSR
n 8000 8000 8000 8000 8000 8000
Bicubic 25.84 121.47 25.57
HE DSM 55.77 4.9920 90.85 16.41 7.737 67.36

SWT & DWT 26.74 22.01 26.45 GC

=1.5 DSM 45.98 9.04 116.8 24.26 11.61 83.4

Proposed technique 30.63 26.01 28.46 SSR DSM 28.61 3.6846 85.50 12.27 4.413 78.15
MSR DSM 5.910 0.3910 20.11 6.771 1.156 13.14

2.4 Undecimated WT based Enhancement Algorithm: MHPF DSM 21.71 7.2688 63.18 12.83 1.546 56.94

In [4] the proposed algorithm uses the undecimated WT 2.6 Algorithm using WT and weighted filter:
which provides higher visual quality improving the local
contrast and resulting in a balanced tonal renedition for the In [7] the proposed algorithm uses a weighted filter for
enhancement of colour images. The enhancement results are enhancing global brightness and contrast of images and WT
shown in fig.[4] while the Visual Contrast Measure(VCM) to enhance the colour information. The flowchart of the
results are given in Table.2. proposed method is shown in fig.3.

125 www.erpublication.org
 Low Contrast Image Enhancement using Wavelet Transform based Algorithms: A Literature Review

Intensi
ty
R,G,B

(b)
Fig 4(a) Original image (b) Enhanced image with pre-processed histogram
equalization
Table 4 shows the comparison of different techniques based
on PSNR & MSE values. The results of PSNR are increasing
Fig.3. Weighted filter algorithm [7]
in pre-processed H.E technique and MSE has decreased.
This paper concludes that the proposed method in [7]
performs better than the auto-level function in the commercial
image-editing software. It not only brightens the darker areas Table 4 Comparisons based on MSE & PSNR [8]
but also preserves the details in highlighted area and removes
noise. Image Techniquename MSE(ForRecov PSNR(dB)(ForRe
(BMPimage ered covered
s) Enhancement) Enhancement)
2.7 Algorithm using non-linear enhancement technique: Hand. Bmp Histogramequal 5.4342e+003 9.6665
(469x700) ization
In [8] the non-linear enhancement technique is used to DWT 3.2720 41.8697
increase the contrast level of an image like WT and Histogram preprocessedhis 8.3631e-005 87.7942
Equalization. In fig 4(a) shows the original image ofhand , togram
Legs. Bmp Histogramequal 1.9219e+003 14.1806
legs & teeth while Fig.4(b) shows the results of pre-processed (338x278) ization
H.E technique. DWT 38.2930 31.1866
preprocessedhis 4.7692e-005 90.2334
togram
Teeth.Bmp Histogramequal 3.2144e+003 11.9469
ization
(275x183) DWT 18.2275 34.4106
preprocessedhis 2.6123e-005 92.8476
togram

2.8 Algorithm using Modern Retinex Technique ( MSR

& SSR):

Paper [9],[10] concludes that Modern Techniques Retinex

(MSR& SSR) performs much better than H.E, Homomorphic
Filtering because it is based on colour consistency theory but
it still suffers from colour violation & unnatural colour
rendition problem which can be improved by applying WT
(a)

III. COMPARISON OF WT BASED IMAGE

ENHANCEMENT ALGORITHM:-
Comparison of all Wavelet transform based all existing algorithm are discussed here.
S.no Reference paper Algorithm Wavelet used Measuring Advantage Disadvantage
parameters
[1] Image Enhancement Directional Mexican Hat MSE Removes interpixel Cannot adapt local
using Wavelets WT+DCT Kernel Wavelet PSNR redundancy &improves information of an
contrast of an image image.
[2] Image Enhancement AHE+DWT 1D DWT Mean()=94.3802 Suitable for low contrast Used only for
using DWT, DCT & +DCT+SVD Standard Deviation() satellite& CT scans greyscale images &
SVD =98.2193 images enhancement not for color images

126 www.erpublication.org
 International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869, Volume-3, Issue-6, June 2015
[3] Resolution DWT+SWT Daubechies PSNR=26.96dB Super resolution images Losses some
enhancement of Wavelet are obtained from low information during
images with resolution images interpolation
interpolation&
DWT-SWT Wavelet
Domain components
[4] Undecimated Undecimated WT UWT VCM=92 Provides high visual
Wavelet transform based enhancement quality & can be used for
based contrast algorithm both colored and _
enhancement greyscale images.

[5] Image resolution & HE+DWT 1-level DWT+ PSNR=26.01dB Provides good resolution Problem of losing
contrast enhancement +SWT+SVD Daubechies 9/7 SSIM=0.9625 & good contrast information in the
using Discrete histogram of the
Wavelet Transform input image.
[6] Contrast DSR+DWT Biorthogonal DSM=56.97 Highly suitable for Problem of
enhancement of dark spline wavelet contrast enhancement of bleaching of image.
images using very dark greyscale
stochastic resonance &colored images
in wavelet domain Better visual perception,
Less computational
complexity.
[7] A new enhancement HE+WT+ weighted WT Enhance high contrast It introduces several
approach for filter images effectively & types of image
enhancing images of also preserves details& artifacts.
digital cameras by removes noise.
changing contrast This method is fully
automatic & requires no
measuring parameter.
[8] Advanced image Preposed HE DWT PSNR=90.2334dB Not only enhances the Used specifically
enhancement based MSE=4.76922-005 image contrast but also for medical images,
on wavelet & preserve original image
histogram quality
equalization for
medical images
[9], a) Colour image Modern techniques WT Provides dynamic range Suffers from Colour
[10] enhancement Retinex(SSR & compression preserve violation
technique: A critical MSR) most of the details &Unnatural colour
review _ rendition

b) Image
enhancement
techniques: A
selected review
and unnatural colour rendition problem. The
The above comparison table concludes that the algorithm undecimatedwavelet transform provides high visual quality
using Directional wavelet andDCT removesinterpixel
redundancy improving contrast of an image but is unable to improving local contrast and resulting in a balanced tonal
adapt local information of an image. Algorithm described in rendition for the enhancement of colour images, so in future
[2] is suitable for low contrast satellite and CT scan images work there is a scope of applying undecimatedwavlet
but is applicable for only grey scale images. Algorithm used transform along with retinex to improve image enhancement
in [3] and [5]provides good resolution and good contrast but results such as colour rendition problem, PSNR.
losses some information . Undecimated transform based
enhancement algorithm provides high visual quality and can REFERENCES
be used for both coloured and greyscale images but losses [1] ShivaniJain ,Jyoti Rani, Image Enhancements using Wavelets ,
some features. DSR along with DWT algorithm is suitable for International Journal of Innovative Research I Computer and
very dark greyscale and coloured images with less Communication Engineering, Vol.2, Isssue 5, May 2014.
computational complexity to provide better visual perception, [2] Aswathy Mohan, Mary Linda PA, ImageEnbhancement using DWT,
DCT, SVD, Int. Journal of Engineering Research and Applications,
but the only problem is bleaching of image. Algorithm used in
Vol.4, Issue 4(Version 1), April 2014, pp.36-46.
[7] enhances high contrast images and also preserves details [3] Mr. G.M. Khaire, Prof. R.P. Shelkikar, Resolution Enhancement of
and removes noise but it still introduces several type of image images with Interpolation and DWT-SWT Wavelet Domain
artifacts.Proposed histogram equalization enhances image Components, International Journal of Application or Innovation in
Engineering & Management(IJAIEM), Vol.2, Issue 9, September 2013.
contrast preserving original image quality but is specifically
[4] NumanUnaldi, SamilTemel&SuleymanDemirci, Undecimated Wavelet
used for medical images. Modern technique retinexi.e. SSR Based Contrast Enhancement, International Journal of Computer,
and MSR in [9], [10] preserves most of the details but suffers Information, Systems and Control Engineering, Vol.7 No:9,2013.
from colour violation and unnatural colour rendition. [5] K.Gunaseelan ,E.Seethalakshmi, Image Resolution and Contrast
Enhancement Using Singular value and Discrete wavelet
decomposition, Journal of Scientific & Industrial Research, Vol.72,
IV. CONCLUSION January 2013, pp.31-35.
Literature review concludes that Modern techniques retinex [6] RajlaxmiChouhan, C. Pradeeep Kumar, Rawnak Kumar and Rajib
Kumar Jha, Contrast Enhancement of Dark Images using Stochastic
(SSR & MSR) performs much better because it is based on
colour consistency theory but it suffers from colour violation

127 www.erpublication.org
 Low Contrast Image Enhancement using Wavelet Transform based Algorithms: A Literature Review

Resonance in Wavelet Domain, International Journal of machine

learning and computing, Vol.2, No.5, October 2012.
[7] Dr.Muna F. Al-Samarie, A New Enhancement Approach for Enhancing
Image of Digital Cameras by Changing the Contrast, International
Journal of Advanced Science and Technology, Vol.32, July 2011.
[8] Yashu Rajput, Vishwashvar Singh Rajput, Anita Thakur, Garima Vyas,
Advanced Image Enhancement Based on Wavelet & Histogram
Equalization for Medical Images, IOSR Journal of Elctronics and
Communication Engineering (IOSRJECE), ISSN:22787-2834 Volume
2, Issue 6(Sep-oct 2012), PP 12-16.
[9] Anish Kumar Vishwakarma, Prof.AgyaMishra , Color Image
Enhancement Technique: A Critical Review, Anish Kumar
Vishwakarma et al / Indian Journal of Computer Science and
Engineering(IJCSE).
[10] Nancy, Er. Sumandeep Kaur, Image Enhancement Techniques: A
Selected Review, IOSR Journal of Computer Engineering (IOSR-JCE)
e-ISSN:2278-0661, p-ISSN:2278-8727 Volume 9, Issue 6( Mar-Apr.
2013), PP 84-88.
[11] P.Suganya, S.Gayathri, N.Mohanapriya, Survey on Image
Enhancement Techniques, International Journal of Computer
Applications Technology and Research, Volume 2-Issue 5, 623-627,
2013.

Namita Naik, M.E. Scholar, Dept. of Electronics &

Telecommunication Engineering ,Jabalpur Engineering college, Jabalpur,
India. And has done Bachelor of Engineering from Vindhya Institute of
Technology and Science, Jabalpur, India.

Dr. Agya Mishra working as Assistant Professor in

Dept. of Electronics & Telecommunication Engineering,, Jabalpur
Engineering College, Jabalpur, India. She has done her Ph.D. and M.Tech
from Maulana Azad National Institute of Technology, Bhopal, India. And
Bachelor of Engineering from Government Engineering College, Ujjain,
India. Her specializations are in the field of Digital Signal & Image
Processing, Higher Order Neural Network, Wavelet transform
applications, computational Intelligence etc.

128 www.erpublication.org