IJOSAT Vol.

2, Issue 2, Oct - Dec 2014

ISSN: 0976-7578 (Online)

A FAST IMAGE SHARPNESS ALGORITHM

IMPLEMENTATION FOR SHARPNESS ENHANCEMENT IN
SATELLITE IMAGES
P.Bhaskar Rao1, T.Ravi Kumar2
Student of M.Tech, Department Of Electronics and Communication Engineering, SISTAM, Srikakulam
2
Assistant Professor, Department Of Electronics and Communication Engineering, SISTAM, Srikakulam
1
bhaskarpagoti53@gmail.com,2ravi.thella@gmail.com
1

ABSTRACT:
Digital Images are playing a vital role in
various fields of Engineering & research.
Analysis of Images which are of Low or
High contrast is the most difficult task.In
this Paper one of the Sharpening technique
for Digital Images is proposed.Sharpening is
one of the most impressive transformations
one can apply to animage. Since it seems to
bring out image detail that was not there
before. What it actually does, however, is to
emphasize edges in the image and make
them easier for the eye to pick out .while the
visual effect is to make the image seem
sharper, no new details are actually
created.The proposed algorithm is referred
to as FISH (Fast Image Sharpness) .It is
named as it has proved to give better results
when compared to the previous Image
sharpening techniques in terms of contrast
and complexity.the inherentinformation
content in the data. It simply emphasizes
certain specified image characteristics.
Enhancement algorithms are generally
interactive and application dependent.
Key WordsBlur, image quality, local
image sharpness, sharpness, wavelet.
I.Image Enhancement Techniques:
Sometimes image obtained from
satellites and conventional and digital

www.ijosat.co.in

cameras lack in contrast and brightness

because of thelimitations of imaging sub
systems and illumination conditions while
capturing image. Images may have different
types of noise. In image enhancement, the
goal is to accentuate certain image features
for subsequentanalysis or for image.
Examplesinclude
contrast
and
edge
enhancement,pseudo-coloring,
noise
filtering, sharpening and magnifying. Image
enhancement is usefulin feature extraction,
image analysis and animage display. The
enhancement processitself does not increase
Some of the image enhancement techniques
are:
1. Contrast stretching
2. Noise filtering
3. Histogram Modification
The paper is discussed as follows
section as follows (ii) Introduction (iii)
Algorithm.(iv)
Local block- based
FISH.(V) Result and discussions(vi)
Conclusion and Results (vii) References
II.Introduction
A useful goal in image processing is
to determine whether one image (region)
appears sharper than another. Algorithms
which can automatically predict perceived
__________________________________________________
International Journal of Science And Technology

IJOSAT Vol. 2, Issue 2, Oct - Dec 2014

sharpness or blurriness are known as

sharpness
estimators
or
blurriness
estimators, respectively. Such algorithms
have been shown to be useful for tasks such
as main-subject detection, image quality
assessment and image restoration.Previous
methods of sharpness/blurriness estimation
have employed a wide variety of
approaches.The vast majority of these
methods operate under the assumption that
the appearance of edges is affected by blur,
and accordingly these methods estimate
sharpness/blurriness by using various edgeappearance models.
For example, Ferzliet al. measure
edge widths in 8x8 blocks, which are then
weighted by a Mean Just-Noticeable Blur
factor employ edge features extracted by
using a Sobel edge detector, and then
combine these features via a circular backpropagation neural network system for blur
estimation. Li et al. compare the kurtoses of
blocks of dominant edge pixels in the input
image with those of a purposely reblurried
version.Other methods have used spectral
information to estimate sharpness.The
authors are with the School of Electrical and
computer Engineering, OklahomaState
University, Stillwater, OK 74078 USA.
Digital Object Identifier 10.1109 /
LSP.20122199980 estimate sharpness based
on the ratio of high-pass to low-pass energy
of the spatial derivative of each line/column.
Sharpness has also been estimated based on
the peakedness of the energy spectrum and
on the uniformity of the energy
spectrum.Various DCT, DWT, and other
transforms have also been used either to
detect edges and/or to model edgeappearance. Sharpness/blurriness has been
estimated based on the kurtosis of DWT
coefficients corresponding to edge blocks,
based on the Lipschitz exponent of the
sharpest edges, based on edge types, and
based on local phase coherence measured

www.ijosat.co.in

ISSN: 0976-7578 (Online)

via complex wavelets. More recently, hybrid

approaches have been developed which
employ a combination of edge-/pixel-based
and transform-based methods. For example,
Chenetalproposed a blur metric that employs
the statistics of the image gradient histogram
and a wavelet-based detail map. Vuetalused
a block-based approach to develop the first
method specifically designed to measure
local sharpness.
Their method estimates the spatial
and spectral sharpness of local image
regions using the slope of the local
magnitude spectrum and the local total
variation; these values are then combined to
generate an image sharpness map. Hybrid
approaches have generally proven to
perform better than edge-only-based or
transform-only-based methods, though at the
expense of added computational complexity.
In this paper, we present a sharpness
estimator, called FISH (Fast Image
Sharpness), which offers the simplicity of a
spectral-based method but with the
improved predictive performance of a
hybrid method. Following from different
international references, FISH operates
under the assumption that perceived
sharpness can be estimated by examining the
energy in high-frequency bands. Here, we
use a three-level separable discrete wavelet
transform (DWT) and measure the logenergy of the DWTsub bands. Sharpness is
estimated based on a weighted geometric
mean of these log-energies. As we will
demonstrate, despite its simplicity, FISH is
competitive with the currently bestperforming techniques. In addition, by
clustering DWT coefficients, we show how
FISH can be easily modified to yield a map
indicating the relative sharpness of each
image region.Thus, unlike most existing
methods FISH can generate sharpness maps.

__________________________________________________
International Journal of Science And Technology

IJOSAT Vol. 2, Issue 2, Oct - Dec 2014

III.Algorithm Implementation:
A. Global Image-Based FISHgiven a
grayscale input image, the FISH algorithm
consists of the following three steps.
1) step-1:
Compute the DWT:The grayscale input
image is decomposed into wavelet sub bands
by using different wavelet filters with n
levels of decomposition. Let denote the LH,
HL, and HH sub bands at DWT level. (The
LL sub band is not used.).where
n=1,2,3.N;
2) Step 2:
Compute the Log-Energy at Each DWT
Level: Images which appear sharp generally
contain more high-frequency content than
images which appear smooth/blurred. To
quantify this effect, we first measure the logenergy of each sub band at each
decomposition level as follows:
(2.a) Where is either or. The quantity is the
number of DWT coefficients in the sub-band
at level. The addition of one is used to
prevent negative values of. Next, we
measure the total log-energy at each
decomposition level via
(2.b) Where the parameter was chosen
empirically to give greater weight to the
energy in the HH sub-band; this band can be
regarded to span a higher radial spatial
frequency (by a factor of) than the LH and
HL bands.
3) Step 3:
Compute the Sharpness Index: Finally, the
three per-level log-energy values , and are
combined as follows to determine a scalar
sharpness index representing theImages
overall sharpness:(3) Here, is the overall
sharpness index; the larger the index, the
greater the perceived sharpness. The factor

www.ijosat.co.in

ISSN: 0976-7578 (Online)

when is used to provide greater weight to the

finer scales (higher-frequency bands).
IV.Local Block-Based Fish:
The previous section described the
FISH algorithm applied to the entire image.
It is also possible to apply the algorithm in a
block-based fashion to determine a map
denoting local perceivedSharpness. To
generate the sharpness map, we compute a
collection of local FISH values using the
DWT coefficients corresponding to each
16x16 block of the image. Following the
procedure describedin, each sub-band is
divided into small blocks of size 8x8, 4x4,
and 2x2 for levels 1, 2, and 3, respectively.
As shown in Fig. 1, the 16x16DWT
scoefficients corresponding to the top-left
16x16 block of the image are assembled by
taking three 8x8 blocks from the level-1
bands, three 4x4blocks

Fig.1.Illustration
of
DWT
coefficients clustering into a 16x16 wavelet
block with 50% overlap to generate the
sharpness map. The orange pixel and its two
adjacent pixels in the sharpness map are
shown according to the orange striped set of
DWT coefficients and two adjacent sets of
DWT coefficients with 50% overlap. Note
that, to promote visibility, the size of the
blocks and sharpness map are not drawn to
scale; the map is 64x64 pixels for a 512x512
input image.From the level-2 bands, and
__________________________________________________
International Journal of Science And Technology

IJOSAT Vol. 2, Issue 2, Oct - Dec 2014

three 2x2 blocks from the level-3 bands.

Equation (3) is then applied to these 16x16
coefficients to compute a FISH index for
this top-left block.
This process is repeated for each
16x16 blocks with 50% overlap between
two consecutive blocks of DWT coefficients
to generate a sharpness map. Because we
use 50% of overlap between neighboring
blocks, each pixel in the sharpness map
corresponds to a block size of 8x8 in the
input image. Fig. 1 (right), illustrates the
sharpness map of the image lena. It is also
possible to collapse the sharpness map into a
scalar sharpness index representing the
images overall sharpness.
This index, , is computed by taking
the root mean square of the 1% largest
values of the local sharpness (FISH) indices
(following from [1]):where denotes the
number of blocks which received the 1%
largest FISH indices of the sharpness map;
and where denotes the FISH indices of these
blocks. The value of 1% is used because, as
argued, the overall perceived sharpness of
an image is largely determined by the
images sharpest regions.
V. Results And Discussion:
Representative results:
The results that demonstrate the
ability of to accurately estimate acrossimage and within image sharpness (only) for
a variety of images containing different
sharpness levels. The images are ordered
based on subjective ratings of sharpness. In
terms of cross-image sharpness, the indices

www.ijosat.co.in

ISSN: 0976-7578 (Online)

generally match the relative perceived

sharpness across these images. For example,
two of the images, petal and zebra, are not
as sharp as images pelicans and branches,
but are clearly much sharper than image
ball. Both FISH and fail to predict the
sharpness of image petal in comparison to
either image airplane (for FISH) or image
zebra (for). We believe that these failure
cases are attributable to the fact that VU
andChandler:

V.B.Fast Wavelet- Based Algorithm

Using Satellite Images:
KALPANA-1:
Kalpana -1 is the First dedicated
meteorological satellite launched by India
Space Research Organization using Polar
Satellite Launch Vehicle on 2002-09-12.
This was the first satellite launched by the
PSLV into the Geo-stationary orbit. The
satellite was originally known as MetSat-1.
On February 5, 2003 it was renamed to
Kalpana-1 by the Indian Prime Minister
AtalBihari Vajpayee in memory of
KalpanaChawlaa NASA astronaut who
perished in the Space Shuttle Columbia
disaster. The METSAT bus was used as the
basis for the Chandrayaan lunar orbiter
mission of 2008.
The satellite features a Very High
Resolution scanning Radiometer (VHRR),
for three-band images and a Data
Relay Transponder(DRT) payload.
VHRR scanning radiometer
The three band images are:
1. Visible
__________________________________________________
International Journal of Science And Technology

IJOSAT Vol. 2, Issue 2, Oct - Dec 2014

ISSN: 0976-7578 (Online)

Apogee
:
kilometers(22,232.0 mi)
Inclination
:
Period
:

2. Thermal Infra red

3.WaterVapour Infra-red
V.C
Insat
Information:

-3d

Satellite

35,779.0
0 degrees
24 hours

Images
(iv)

Instruments

Mission Type

weather

VHHR

Operator

ISRO

Images are followed by HUDUD Toofan

Website

ISROWeb-site

Mission Duration

7 years

(i)

Spacecraft Properties

Manufacturer
:
ISRO satellite
center ,Space Application center
Launch mass
:
kilograms (2,340 lb)

1060

Power

550 watts

(ii)

Start of Mission

SAT:INSAT-3D ING
12-10-2014/10:30 GMT
Thermal infrared 2count 12.12-10-2014/16:00 IST
LIC SECTOR ASIA_MER MERCATOR
FIG:2:Thermal infrared band image with 10.8um

Launch Date
:
2002 10:24:00 UTC

12 September

Rocket

PSLV C4

Launch site
FLP

satishDhawan

(iii)

Orbital Parameters

Reference system
:
Regime
:
Longitude
:
Eccentricity
:
Prigee
:
kilometers (22,249.9 mi)

www.ijosat.co.in

Geocentric
Geostationary
740 East
0.00um
35,807.7

FIG:3: IMG VIS 0.6um (R) IMG_VIS 0.65um(G)

/IMG_TIRI 10.8um
SAT:INSAT-3D ING
12-10-2014/10:30 GMT
Thermal infrared 2 count 12.0um 12-10-2014/16:00IST
LIC SECTOR ASIA_MER MERCATOR

__________________________________________________
International Journal of Science And Technology

IJOSAT Vol. 2, Issue 2, Oct - Dec 2014

Enter the percent of Pixels With Noise : 10

n=0

ISSN: 0976-7578 (Online)

sharpness or bluriness that happened in

image.It offers the simplicity of spectralbased model.
Real time satellite images
can also be used to detect the effect of
cyclones and all drastic changes in
environment.
References:

Fig 4: Output With Fish Algorithm

1.
Phong V. Vu and Damon M.
Chandler.
2.
C. T. Vu, T. D. Phan, and D. M.
Chandler, : A spectral and spatialmeasure
of local perceived sharpness in natural
images, IEEE Trans.Image Process., vol.
21, no. 3, Mar. 2012.
3.
P.Marziliano, F.Dufaux, S.Winkler,
and
T.
Ebrahimi,
A
noreferenceperceptual blur metric, in IEEE
Int. Conf. Image Processing, ICIP2002,
2002, vol. 3.
4.
R. Ferzli and L. J. Karam, A noreference objective image sharpnessmetric
based on the notion of just noticeable blur
(JNB), IEEE Trans.Image Process., vol.
18, no. 4, Apr. 2009.

Fig 5: Result with Block Based Fish

Algorithm

5.
N. D. Narvekar and L. J. Karam, A
no-reference perceptual imagesharpness
metric based on a cumulative probability
of blur detection,inInt. Workshop on
Quality of Multimedia Experience,
QoMEx, Jul.2009.

VI. Conclusion:
In fast image sharpness algorithm
implementation
(FISH)for
sharpness
enhancement in digital images.From this
technique ,we can perform impressive
transformation to apply an image .If to
emphasize edges in images and easier for
eye pick .It can automatically estimates the

www.ijosat.co.in

6.
H. Liu, J.Wang, J.Redi, P. LeCallet,
and I. Heynderickx, An efficientnoreference metric for perceived blur, in
Eur. Workshop on VisualInformation
Processing, EUVIP 2011, Jul. 2011.

__________________________________________________
International Journal of Science And Technology