View
212
Download
0
Category
Preview:
DESCRIPTION
.
Citation preview
Asian Transactions on Fundamentals of Electronics, Communication & Multimedia (ATFECM) (ATFECM ISSN: 2221-4305) Volume 01 Issue 05
Nov 2011 ATFECM-30109054Asian-Transactions 1
Abstract A well known image compression algorithm named Set Partitioning in Hierarchical Trees (SPIHT) is an efficient
wavelet-based progressive image-compression technique,
designed to minimize the mean-squared error (MSE) between the
original and decoded imagery. Since wireless channels suffer from
significant bit error rates, some mechanism to protect the encoded
image required. Without such a mechanism, the channel bit
errors will prevent accurate decoding of the image. Many error
control solution have been proposed to address the problem of
transmitting images over wireless channels. These error control
coding (channel coding) techniques add redundancy at the source
level. Asymmetric modulation method, Hierarchical Quadrature
Amplitude Modulation (HQAM) provides alternative means of
Unequal error protection to protect the transmitted sensitive
coded bits without increasing the bandwidth. The paper proved
that HQAM method can reduced sensitive bit error when using
SPHIT image encoding where SPHIT has a drawbacks on error
resilience. The performance of SPIHT coded image transmission
is evaluated using gray test image and for different values of the
modulation parameter and different values of bpp (bit-per-pixel).
Index Terms Error protection, HQAM, Image compression,
SPIHT Image coding, Wireless channel.
I. INTRODUCTION
any techniques have been proposed for the transmission
of images over wireless networks. One obvious
technique is to take an existing image encoder, and then
protect the bit stream using convolution codes a separate
source-channel coding scheme. New algorithms for image
compression based on wavelets have been recently developed
[1-4]. These methods have resulted in practical advances such
as: superior low-bit rate performance, continuous-tone and bit-
level compression, lossless and lossy compression, progressive
Manuscript received October 7, 2011. This work was supported by the
Ministry of Science, Technology and Innovation (MOSTI) under grant 01-01-
15-SF0115.
Md. Abdul Kader is with the School of Computer and Communication
Engineering, University Malaysia Perlis, Malaysia. (e-mail:
kdr2k4@yahoo.com).
Prof. Dr. Farid Ghani is with the School of Computer and Communication
Engineering, University Malaysia Perlis, Malaysia. (e-
mail:faridghani@unimap.edu.my).
Assoc. Prof. R. Badlishah Ahmed is with the School of Computer and
Communication Engineering, University Malaysia Perlis, Malaysia. (e-mail:
badli@@unimap.edu.my).
transmission by pixel accuracy and resolution, region-of-
interest coding and others. One of the most efficient
procedures that fulfil the above goals is the Set Partitioning in Hierarchical Trees (SPIHT) algorithm [1]. It was introduced
by Said and Pearlman [1, 5]. Some of the best results are
highest PSNR values for given compression ratios and for a
wide variety of images have been obtained with SPIHT.
Consequently, it is probably the most widely used wavelet-
based algorithm for image compression, providing a basic
standard of comparison for all subsequent algorithms.
Although the SPIHT coder is a progressive encoding in which
the decoding can be terminated at any bit in the stream, it has
very poor error resilience (single bit errors often cause a loss
of synchronization between the encoder and decoder and can
result in a badly distorted image). Therefore, if channel
included bit errors, decoding can be terminated before the bit
containing the error is decided. The error is likely to cause
later bits to be decided incorrectly as well, which significantly
increases decoding error.
Hierarchical modulation methods provide alternative ways
for unequal error protection (UEP) to the transmitted bits
without increasing the bandwidth. The most commonly used
hierarchical modulation method is Hierarchical Quadrature
Amplitude Modulation (HQAM). This modulation technique
assigns unequal protection to all the transmitted bits, hence is
classified as unequal error protection (UEP) method of
modulation [6]. HQAM technique is a modification of
Quadrature Amplitude Modulation (QAM) technique where
QAM provides equal error protection (EEP) to all the
transmitted bits. Thus UEP modulation methods are to be
preferred particularly for coded image data transmission, as
these will allow gradual protection of the data with regard to
its importance i.e. more important data will get a stronger
protection than less important. This is a simple and efficient
approach in which non-uniform signal-constellation is used to
give different degrees of protection. The advantage of this
method is that different degrees of protection are achieved
without an increase in bandwidth; in contrast to channel
coding that increases the data rate by adding redundancy [7-
11].
In this paper an image transmission system has been
proposed where Set Partitioning in Hierarchical Tree (SPIHT)
Unequal Error Protection for SPIHT Coded
Image Transmission over Erroneous Wireless
Channels
Md. Abdul Kader, Farid Ghani and R. Badlishah Ahmed
M
Asian Transactions on Fundamentals of Electronics, Communication & Multimedia (ATFECM) (ATFECM ISSN: 2221-4305) Volume 01 Issue 05
Nov 2011 ATFECM-30109054Asian-Transactions 2
algorithm is used as an image coder and 16-HQAM technique
is used for transmission of the coded image over wireless
erroneous (AWGN and Fading) channels. PSNR analysis is
presented for the proposed image transmission system and the
performance comparisons are carried out through computer
simulation with gray image as test image for different values of
modulation parameter and different value of bpp (bit-per-
pixel).
This paper has been organized as follows: In section II,
general model of image transmission system is given. Section
III considers an overview of SPIHT image compression
technique. In section IV overview of Hierarchical 16-QAM
describes. Based on computer simulation results, the
performance of HQAM for the transmission of SPIHT coded
images is considered in Section V.
II. MODEL OF IMAGE TRANSMISSION SYSTEM
The essentials of the image transmission system considered
here are shown in Fig. 1. The source encoder encodes the
source image using appropriate image compression technique.
For the protection of coded image in Fig. 1 channel encoder
add redundancy to the coded image by using appropriate
channel coding technique. Modulator modulates the coded
image and transmits through wireless channel. QAM is
invariably used as the modulation technique [13, 14]. The
channel introduces noise and distortion to the transmitted
image. The demodulator receives the image data with error
and demodulates it. After channel decoding, the coded image
is decompressed.
Fig. 1. Model of image transmission system
III. OVERVIEW OF SPIHT ALGORITHM
SPIHT stands for Set Partitioning in Hierarchical Trees. The
term hierarchical trees refers to the quadtrees. Set partitioning
refers to the way these quadtrees partition the wavelet
transform values at a given threshold.
The SPIHT algorithm is based on hierarchical set
partitioning, which can be thought of as a divide-and-conquer
strategy. The SPIHT algorithm views wavelet coefficients as a
collection of spatial orientation trees, with each tree consisting
of coefficients from all subbands that correspond to the same
spatial location in an image. A partitioning rule is used to
divide a given set into smaller subsets so that significant
coefficient can be efficiently isolated.
The SPIHT consists of two main stages namely sorting and
refinement. For practical implementation, SPIHT maintains
three linked lists viz. the list of insignificant pixels (LIP), the
list of significant pixels (LSP) and the list of insignificant sets
(LIS). At the initialization stage, SPIHT initializes the LIP
with all the pixels in the highest level of the pyramid (i.e. LL
subband), the LIS with all the coefficients at the highest level
of the pyramid except those, which don't have descendents,
and LSP as an empty set. During the sorting pass, the
algorithm first traverses through the LIP, testing the magnitude
of its elements against the current threshold and representing
their significance by 0 or 1. Whenever a coefficient is found
significant, its sign is coded and it is moved to LSP. The
algorithm then examines the LIS and performs a magnitude
check on all coefficient of set. If a particular tree/set is found
to be significant, it is partitioned into its subsets (children and
grandchildren) and tested for significance. Otherwise a single
bit is appended to the bit stream to indicate an insignificant set
(or zero-tree). After each sorting pass SPIHT outputs
refinement bits at the current level of bit significance of those
pixels which had been moved to LSP at higher threshold,
resulting in the refinement of significant pixels with bits that
reduce maximum error. This process continues by decreasing
current threshold by factor of two until desired bit rate is
achieved.
The SPIHT coder generates different type of bits that have
different degree of vulnerability to the errors. The effect of
error in some bits is more severe, damaging the image globally
by disturbing the synchronization between the encoder and
decoder. On the other hand, some bits are less sensitive to
errors, and damage the image locally without disturbing the
synchronization.
Based on the degree of their sensitivity, the bits generated
by SPIHT algorithm can be classified into two classes Critical
Bits (CB) and Non-Critical Bits (NCB). The critical bits are
those, which causes the loss of synchronization between the
encoder and decoder. A single bit error in critical bit causes
the failure of the reconstruction process after that point. It
consists of significant bits generated during LIP and LIS tests.
The non-critical bits on the other hand cause less severe errors.
The effect of error in a noncritical bit is limited to a single
coefficient and does not disturb the progression of the
decoding process after it occurs. The non-critical bits consist
of the sign and refinement bits.
IV. HIERARCHICAL 16-QAM
Hierarchical Quadrature Amplitude Modulation (HQAM) is
more spectrally efficient and dc-free modulation scheme.
Hierarchical transmission system is composed of a hierarchical
source coder and the corresponding channel coder divides the
information into several layers according to their significance,
and transmits each layer with different reliability according to
the layers.
Fading and
Noise Channel
with Errors
Received Data
with Error
Compressed
Data
Transmitted
Image
Channel
Encoder Modulator
Source
Encoder
Compression
De-
Modulator
Channel
Decoder Received
Image
Source
Decoder
De-Compression
Asian Transactions on Fundamentals of Electronics, Communication & Multimedia (ATFECM) (ATFECM ISSN: 2221-4305) Volume 01 Issue 05
Nov 2011 ATFECM-30109054Asian-Transactions 3
The basic idea behind hierarchical modulation consists of
the partitioning the coded image data stream into two parts: the
significant or high-priority (HP) information and the non-
significant/refinement or low-priority (LP) information.
After channel encoding, the HP and LP information are de-
multiplexed into a single stream and mapped on non-uniformly
spaced constellation points creating different levels of error
protection [12]. In this way, HQAM provides a way of
Unequal Error Protection (UEP) to the transmitted data bits, in
which the high priority data bits (HP) of the coded image are
mapped to the Most Significant Bits (MSB) of the modulation
constellation points while the low priority data bits (LP) of the
coded image are mapped to the Least Significant Bits (LSB) of
the modulation constellation points [10]. Unequal error
protection (UEP) modulation methods are to be preferred
particularly for image data transmission, as these will allow
gradual protection to the transmitted data with regard to its
importance.
Using HQAM will, therefore, result in improved image
quality at low channel signal to noise ratio (SNR) conditions,
since the highly sensitive HP data bits are mapped to the
MSBs of the HQAM with low bit error rate (BER). However,
for the sake of simplicity only 16-HQAM is considered in this
paper.
In Hierarchical QAM, it is possible to give the higher
protection to the most important data (significant bits) by
changing the value of modulation parameter . is the ratio of
the distance b between quadrants to the distance c between the
points within a quadrant. Referring to Fig. 2(a) the modulation
parameter = b/c. For a given transmitted signal power the
sum of b/2 and c should remain constant. The value of
should not exceed the square root of the carrier power pc.
Otherwise, the constellation points of the same quadrant will
overlap. Fig. 2(a), (b) and (c) shows the constellation diagram
for = 3, = 2 and = 1 respectively.
(a) = 3
(b) = 2
(c) = 1
Fig. 2. Constellation diagrams of 16-HQAM
Referring to Fig. 2(c), the minimum distance of all
constellation point is equal (2d=b=c, i.e. HQAM results in
QAM) and the two MSB represent the HP bits which have
lower BER than the two LSB bits. LSB bits are representing
LP bits. It can be seen that the four symbols in every quadrant
have the same HP bits but different LP bits; this is also called
constellation overlapping which ensures the HP bits to be
transmitted correctly [14, 22].
Higher portion of the transmitter power means better
protection against the channel errors than the lower portion of
the power. This is a simple unequal error protection without
introducing any redundancy to the modulated data. In this case
the performance of the HP will be improved at the expense of
LP. However, by increasing the degree of non-uniformity, (b >
c) the improvement of the HP performance is significant, at the
expense of the LP sub-channel.
1001 1011 0001 0011
1000 1010 0000 0010
1101 1111 0101 0111
1101 1111 0100 0110
10
5
-5
-10
Q
-10 -5 5 10
I
1001 1011 0001 0011
1000 1010 0000 0010
1101 1111 0101 0111
1101 1111 0100 0110
Q 10
6
-6
-10
c b = b/c,
Asian Transactions on Fundamentals of Electronics, Communication & Multimedia (ATFECM) (ATFECM ISSN: 2221-4305) Volume 01 Issue 05
Nov 2011 ATFECM-30109054Asian-Transactions 4
V. SIMULATIONS AND RESULTS
A simulation for SPIHT coded image transmission and
reception was carried out using 16-HQAM technique. In this
simulation 512X512 gray-scale Lena image is used as test
image with 9-level wavelet decomposition. Biorthogonal
(bior4.4) wavelet is used as wavelet basis function. The flow
diagram of the proposed image transmission system simulation
is shown in Fig. 3. It has been simulated in MATLAB 7.8
(R2009a). In proposed simulation, source image is
decomposed by Biorthogonal wavelet basis function. SPIHT
encoding technique is applied to the wavelet decomposed
image. After encoding the generated bit stream is partitioned
into two levels based on their sensitivity. Higher sensitive data
bits (e.g. image size, number of bit plane, decomposition level
etc.) are mapped to the HP bits and lower sensitive data bits
are mapped to the LP bits in 16-HQAM modulation
constellation points as shown in Fig. 2(c). After mapping with
the constellation points, the generated signals are transmitted
over wireless erroneous (AWGN and fading) channel. The
received signals with noise are demodulated and merge back to
a single bit stream. SPIHT decoding is applied to the bit
stream. After inverse wavelet decomposition, image data are
written into a graphic file.
Fig. 3. Proposed simulation flow diagram of SPIHT coded image transmission
and reception using 16-HQAM technique
The results have been evaluated by peak-signal-to-noise-
ratio (PSNR) of the received reconstructed image. The PSNR
of an image can be calculated using (1)
MSE
KPSNR 10log10
(1) (1)
y x
yxgyxgNM
MSE2' ,,
1
where, g(x,y) and g(x,y) represent the gray values of the pixels
in the original image and the reconstructed image respectively.
M and N represent the width and height of the image
respectively, while K is the maximal gray value of the image
[15].
This simulation calculates the PSNR value of the received
reconstructed image for different bit-rate (0.1 to 1.0) and also
for different values of modulation parameter (1 to 5) through
AWGN channel and noisy fading channel. PSNR versus Bit-
rate (bpp) curves are shown for the transmission of SPIHT
coded gray-scale image using 16-HQAM technique over
AWGN channel in Fig. 4 and over noisy fading channel in Fig.
5. From Fig. 4, it is shown that for the lower value of , PSNR
vs. bpp curves are not smooth because of channel noise but the
PSNR of the reconstructed image increases when the value of
increases for a fixed bit-rate. Fig. 5 also shows the same
characteristics as Fig. 4 for noisy fading channel.
For the transmission of SPIHT coded images, the SNR was
kept at a fixed value of 18 dB. The results are shown in Fig. 6
for AWGN channel and in Fig. 7 for noisy fading channel
Fig. 4. PSNR vs. Bit-rate graph for different values of modulation parameter
for the transmission over AWGN channel
Fig. 5. PSNR vs. Bit-rate graph for different values of modulation parameter
and transmission over noisy fading channel
HP LP
Transmitted
Image
Read Image data
from a Graphic File
Wavelet
Decomposition
SPIHT
Encoding
Bit Stream
Partitioning
16-HQAM
Modulation
MSBs
LSBs
Signal
Transmission
SPIHT
Decoding
Bit Stream
Merging
16-HQAM
Demodulation
MSBs
LSBs
Signal
Reception
HP LP
Write Image data to
a Graphic File
Inverse Wavelet
Decomposition
Received
Image
Add Noise
with
Transmitted
Signal
Asian Transactions on Fundamentals of Electronics, Communication & Multimedia (ATFECM) (ATFECM ISSN: 2221-4305) Volume 01 Issue 05
Nov 2011 ATFECM-30109054Asian-Transactions 5
where each row represents the reconstructed images for the
fixed value of bit-rate (bpp) and different values of modulation
parameter, and each column represents the reconstructed
images for the fixed value of modulation parameter and
different values of bit-rate. It can be seen that for a fixed value
of bit-rate when the value of increases the quality of the
reconstructed images also improves. For example in Fig. 6,
when bit-rate is 0.2 bpp and =1, there are too much distortion
occurs in the reconstructed image for channels noise.
If increases then the amount of distortion decreases and
gradually increases the quality of the reconstructed images. In
Fig. 7, for a fixed bpp when the value of increases the effects
of fading in the reconstructed images decrease.
VI. CONCLUSIONS
In this paper an image transmission system is proposed
where Hierarchical 16-QAM technique is used to get unequal
error protection for the transmission of SPIHT coded image
over erroneous wireless channels. The advantage of this
system is that unequal error protection is achieved without an
increase in bandwidth in contrast to channel coding that
increases the data rate by adding redundancy to the transmitted
signal. The SPIHT coded image is divided into two sub-
streams (HP and LP) based on their sensitivity and transmits
using 16-HQAM technique through erroneous wireless
channel. Thus the sub-streams obtain unequal error protection
without increasing the bandwidth. Simulation results shows
that the proposed image transmission system can improve
PSNR of the reconstructed images for lower channel SNR
using 16-HQAM technique without any additional hardware.
REFERENCES
[1] Said, A. and Pearlman, W.A., A new, fast, and efficient image codec
based on set partitioning in hierarchical trees, IEEE Trans. on Circuits
and Systems for Video Technology, 6(3), 243250, 1996.
[2] J.M. Shapiro, Embedded image coding using zero-trees of wavelet coefficients, IEEE Transactions Signal Processing, vol. 41, Dec. 1993.
[3] D. Taubman, High Performance scalable image compression with
EBCOT, IEEE Transactions on Image Processing, vol. 9, July 2000.
[4] I. Hontsch and L. Karan Locally adaptive perceptual image coding,
IEEE Transactions on Image Processing, vol. 9, September 2000.
Bit-rate (bpp) =1 =2 =3 =4 =5
0.2
0.4
0.6
0.8
1.0
Fig. 6. Reconstructed images using different bit-rate (bpp) and different value of modulation parameter for AWGN channel
Asian Transactions on Fundamentals of Electronics, Communication & Multimedia (ATFECM) (ATFECM ISSN: 2221-4305) Volume 01 Issue 05
Nov 2011 ATFECM-30109054Asian-Transactions 6
Bit-rate =1 =2 =3 =4 =5
0.2
0.4
0.6
0.8
1.0
Fig. 7. Reconstructed images using different bit-rate (bpp) and different value of modulation parameter for erroneous fading channel
[5] Said, A. and Pearlman,W.A., Image compression using the spatial-
orientation tree, IEEE Int. Symp. on Circuits and Systems, Chicago, IL,
279282, 1993.
[6] Yoong Choon Chang; Sze Wei Lee; Komiya, R.; , "A low-complexity
unequal error protection of H.264/AVC video using adaptive
hierarchical QAM," Consumer Electronics, IEEE Transactions on ,
vol.52, no.4, pp.1153-1158, Nov. 2006.
[7] M. Mahdi Ghandi and M. Ghanbari, "Layered H.264 video transmission
with hierarchical QAM," Elsevier J. Visual Commun. Image
Representation, Special issue on H.264/AVC, vol. 17, no. 2, pp. 451 -
466, April 2006.
[8] B. Barmada, E.V. Jones, Adaptive mapping and priority assignment for
OFDM, Proceedings of the IEE Conference of 3G Mobile
Communication Technologies, London, 2002.
[9] Seamus OLeary, Hierarchical transmission and COFDM systems,
IEEE Transactions on Brod., vol. 43, no. 2, pp. 166-174, June 1997.
[10] Chee-Siong Lee, Thoandmas Keller and Lajos Hanzo, OFDM-Based
turbo-coded hierarchical and non-hierarchical terrestrial mobile digital
video broadcasting, IEEE Transactions on Broadcasting, vol. 46, no.1,
pp. 1-22, March 2000.
[11] B. Barmada, M. M. Ghandi, E. V. Jones, M. Ghanbari, Prioritized
transmission of data partitioned H.264 video with hierarchical QAM,
IEEE Signal Processing Letters, v.12. no.8, pp.577-580, August 2005.
[12] Sadough, S.M.S.; Duhamel, P.; , "On the Interaction Between Channel
Coding and Hierarchical Modulation," Communications, 2009. ICC '09.
IEEE International Conference on , vol., no., pp.1-5, 14-18 June 2009.
[13] W.K.Pratt, Digital Image Processing (New York: Wiley 1978).
[14] L. Hanzo, W. Webb, T. Keller, Single and multi carrier QAM: principles
and applications for personal communications, WLANs and
broadcasting, John Wiley and Sons, 2000.
[15] Yong Liu; Ai-dong Men; Zi-yi Quan; Bo Yang; , "Unequal Error
Protection for SPIHT-Coded Image Transmission Based on MIMO
Adaptive Channel Assignment Policy," Advanced Computer Control,
2009. ICACC '09. vol., no., pp.781-785, 22-24 Jan. 2009.
Recommended