Journal of Signal and Information Processing, 2011, 2, 316-321
doi:10.4236/jsip.2011.24045 Published Online November 2011 (http://www.SciRP.org/journal/jsip)
1
Fast Encoding-Decoding of 3D Hyperspectral
Images Using a Non-Supervised Multimodal
Compression Scheme
Mourad Lahdir1, Amine Nait-ali2*, Soltane Ameur1
1Laboratoire d’Analyse et de Modélisation des Phénomènes Aléatoires (LAMPA), Département d’Electronique, Faculté Génie
Ele ctri que et Informatique, Université Mouloud Mammeri, Tizi-Ouzo u, Algérie; 2Laboratoire Images, Signaux et Systèmes Intelligents
(LiSSi), Université Paris-Est Créteil, Créteil, France.
Received June 8th, 2011; revised August 14th, 2011; accepted August 26th, 2011.
ABSTRACT
We introduce in this paper an extension of the Multimodal Co mpression technique (MC) for the purpose of coding hy-
perspectral image sequences. The main idea requires few steps, namely: 1) reducing the size of the sequence by insert-
ing smooth images con taining less information in to the remaining images o f the same sequence, 2) then cod ing the new
compacted sequence using 3D-SPIHT algorithm. In this new scheme, called MC-3D-SPIHT, the insertion is achieved
only in the contour of each image, according to a non-sup ervised way, so that one can preserve the Region of Interest
(ROI) quality. For this purpose, a mixing function is employed. After the decoding process, inserted images are ex-
tracted by a separation function and the original sequence is reconstructed. By considering data from AVIRIS database,
we will show how one decrease significantly the computing time for bo th coding and decoding.
Keywords: Multimodal Compression, Hyperspectral Image s, 3D-SPIHT
1. Introduction
Hyperspectral images provide finer spectral information
then traditional multispectral images. However the volume
of generated data is dramatically huge. Consequently, data
compression becomes essential for economical distribution
when spaceborn hyperspectral data are regularly available.
The term hyperspectral is generally used for spectral data
containing hundreds of samples of spectra. The hyperspec-
tral image s thus pre sent specific characteris tics that r equire
to be exploited by some specific compression algorithms
[1]. Since hyperspectral sequence images consider a set of
images, they can be regarded somehow as volumetric data
requiring specific techniques of compression.
Based on the techniques available in the literature, var-
ious algorithms and standards have been developed to deal
with this type of data. For instance, wave let transform has
been efficiently used for 2D image coding [2,3]. Besides,
it is considered as the kernel o f the s tandard JPEG2000 [4].
Extended to volumetric images, JPEG2000 standard has
been widely applied to 3D hyperspectral images encoding
[5]. Afterwards, the 3D wavelet transformation has been
efficiently employed for various types of data through [6],
[7] and [8]. Recently, a 3D anisotropic wavelet decompos-
ition which includes an adaptation of the zerotree structure
[1] highlighted the potential of using such scheme to com-
press 3D hyperspectral images.
In this paper, we propose a new approach to pre-pro-
cess 3D hyperspectral images before using any compres-
sion scheme. The idea consists in compacting (i.e. redu-
cing the number of images of a sequence to compress)
any volume/sequence in a context of Multimodal Com-
pression based on the con cept introduc ed in [9] related to
image-signal merging and video-signal merging of bio-
medical data. The scheme presented in this work is con-
sidered as a variant and an extension since it deals with
3D images.
Generally speaking, the id ea of MC consists in merging
data using an insertion function (non-supervised scheme)
into an image or a set of images, before the encoding pr-
ocess. Afterwards, a separation functio n is used to extract
the required inf ormation from the decoded data.
In this work, some selected images from 3D hypersp-
ectral images are inserted in the remaining images of the
same sequence. This produces a compacted volume com-
pressed using a 3D-SPIHT algorithm which outperforms

Fast Encoding-Decoding of 3D Hyperspectral Images Using a Non-Supervised Multimodal Compression Scheme317
the CCSDS and JPEG 2000 standards. Consequently, a
fast encoding/decoding is achieved without any signifi-
cant loss of information.
This paper is organized as follows: in Section II, the
methodology of Multimodal Compression extended to
hyperspectral volumetric data, is presented. Results and
performance analysis evaluated on AVIRIS database are
presented in section III. Finally, a conclusion is provided
in section IV.
2. Methodology
We consider an hyperspectral image sequence, denoted
by iwhere i = 1,···,K refers to channels. Each channel
corresponds to an
S
M
N image.
The Multimodal Compression of this sequence re-
quires various phases, namely, 1) analysis, 2) insertion
and 3) encod ing. The process is inverted for the decoding
purpose (see Figure 1).
2.1. Analysis Phase
In this phase, hyperspectral images to be compressed are
sorted so that those containing less information are con-
sidered potentially appropriate to be merged in the re-
maining images. For this purpose, an objective criterion
should be defined. This criterion can perform a simple
statistical analysis along the image ch annels. In this work,
we consider that the smooth images are relatively poor in
terms of information. Therefore, the variance is used as
indicator to sort images from the highest variance to the
lowest one. Consequently, the number of images L that
can potentially be merged depends on the global size of
the Region of Interests in the whole sequence.


global ROI
LNN MN

 
 
K
(1)
where global and NMN
R
OI is to the number
of pixels corresponding only to the Region of Interests in
N
Figure 1. Multimodal Compression scheme applied to hy-
perspectral images.
the sequence to be compressed. is the floor func-
tion. 

2.2. Mixing Phase (Non-Supervised Scheme)
Definition: we call a non-supervised scheme of mixing
function, a procedure which consists in replacing some
pixels of a host image by other useful pixels provided by
another source [10] according a rule defined by the user
and matchin g t he ap pl i c at i o n.
In this case, we consider that L smooth images have
been selected. Corresponding pixels are interleaved in
the contours of the
K
L

remaining images after a
down-sampling process. In other words, in the interleav-
ing process, each pixel over two pixels that belong to the
ROIN, is replaced by another one that belong to the
smooth images to be embedded, as specified in Figure 2.
In such a case, we define two regions, namely, 1) Region
of Interest (ROI) and 2) Region of Insertion (ROIN).
Only ROINs should be down-sampled since we consider
that central regions (ROIs) contain the main information
that should not be distorted. As it is shown in Figure 2,
the central region which forms an ROI for the compres-
sion phase is left without sampling.
In order to reduce the size of the volumetric image to
be compressed, L images should be embedded within
K
L

images by considering the following condition:
0LK2


In the extreme case (i.e. 2LK) no ROIs are used
and the volumetric image to be compressed is exactly
half the initial size.
If the size of the ROI in each image is 11
M
N

, the
number of samples (called here, the capacity) that could
be embedded in each image will be given by:
11
2
M
NM N
C   (2)
Hence, knowing the capacity of insertion, the number
of images that could be dispatched is given by:
ins
K
C
K
M
N



(3)
Finally, the size of volumetric image is reduced to:
ins
SKK

 (4)
2.3. Encoding Phase
In this phase, the reduced volume is compressed using
the SPIHT algorithm. This encoder which is wave-
let-based has been largely employed to compress 1D, 2D
data. Afterwards, it has been extended for volumetric
images (3D-SPIHT) which takes into account pixel cor-
relations along different resolutions as shown in Figure 3.
Moreover, this type of codec is suited for progressive

Fast Encoding-Decoding of 3D Hyperspectral Images Using a Non-Supervised Multimodal Compression Scheme 318
Figure 2. Compacting process. Pixels interleaving in the
mixing phase.
Figure 3. 3D-SPIHT arbre structure.
encoding, regarded as an imp ortant functionality.
Since this algorithm is well known, it will not be de-
scribed in this paper. For this purpose, detailed informa-
tion can be found in [9- 11].
2.4. Decoding Phase
One of the 3D-SPIHT properties is that the algorithm
allows progressive decoding of codestreams. At this ph-
ase, a reduced volume containing the data mixture is ob-
tained.
2.5. Separation Phase
After the decoding phase, this step consists in extracting
the embedded images from the contours of each image
corresponding to a decoded reduced volume. Afterwards,
an interpolation is performed on the contours in order to
estimate missing pixels values. The interpolation used
here is linear. It is calculated from the neighbourhood of
each missing pixel. Hence, the interpoled pixel pi(m,n) is
given by:
 


 
2
1
1
,,11
4
ll
il
pmnpmnpm n


,
(5)
where (m,n) defines the int erpoled pixel position.
3. Simulation Results
To evaluate the performance of a Multimodal Com-
pression scheme using 3D-SPIHT on a sequence of hy-
perspectral images, experiences have been performed
according to the following three phases:
1) Comparison phase: 3D-SPIHT is compared to stan-
dards such as: CCSDS (The Consultative Committee for
Space Data Systems) [12] and JPEG 2000 [4,6]. The aim
of this phase is to show the superiority of 3D based-en-
coders, namely 3D-SPIHT, compared to 2D based-en-
coders.
2) Analysis phase.
3) Multimodal Compression based 3D-SPIHT: the
3D-SPIHT is included in the context of Multimodal
Compression as described in Section 2.
For this purpose, we have used several hyperspectral
sequences from AVIRIS database (Airborne Visible In-
frared Imaging spectrometer). We have used a dataset of
the Yellowstone scene, acquired in 2006 and having a
size of 512 × 614 over 224 optical bands. This AVIRIS
calibrated radiance images can be downloaded from
http://aviris.jpl.nasa.gov/html/aviris.freedata.html.
Comparison phase
As evoked above, 3D-SPIHT is compared in terms of
bit-distortion to both CCSDS standard and JPEG 2000.
For this purpose, a sequence of 16 images has been used.
Since these coders are wavelet-based, three levels of
decomposition have been considered using bi-orthogonal
filters 9/7, as recommended by the CCSDS. Simulations
were performed using the software TER 2.02, which is
an implementation of the recommendations of CCSDS
image compression (Recommended Standard CCSDS
122.0-B-1 Blue Book). http://gici.uab.es/TER/
For JPEG2000 coder, we have used Kakadu Version
5.11 which implements the Part 1 and the Part 2 of
JPEG2000 standard.
For a range [0.25 - 2] bpppb (bits per pixel per band),
the averaged Peak Signal-to-noise, Ratio () is
calculated, where is commonly defined by: A
PSNR
PSNR

2
16
10
21
10logPSNR MSE








(6)
where MSE is the mean square error between the original
and reconstructed image.
As it can be shown on Figure 4, the performance cur-
ves show that the averaged increases with the
bit-rate (bpppb) according to a law which can be appro-
ximately logarithmic. On the other hand, one can point
out that 3D-SPIHT outperforms, within the analyzed
range, both JPEG2000 and CCSDS. This was somehow
expected since 3D-SPIHT takes into account the correla-
PSNR

Fast Encoding-Decoding of 3D Hyperspectral Images Using a Non-Supervised Multimodal Compression Scheme319
0.2 0.40.6 0.811.2 1.4 1.6 1.8 2
40
45
50
55
60
65
70
Bit-rate(bpppb)
PSNR
A
(dB)
CCSDS
3D-SPI HT
J PE G 2000
Figure 4. Comparison between CCSDS, JPEG 2000 and
3D-SPHIT in terms of bite-rate distortion curve. Results
show that the 3D-SPHIT outperforms the other codecs.
tion along the sequence i mages to be compressed.
Based on this result, the Multimodal Compression
scheme will integrate 3D-SPIHT for the encoding and
decoding purpose.
3.1. Analysis Phase
In this phase, the initial hyperspectral sequence i is
analyzed in terms of statistics so that one can determine
the images that potentially can be embedded into other
images. For this purpose, using a sequence of 36 images
extracted of the Aviris basis acquired on Yellowstone
WY in 2006, the variance of each channel has been cal-
culating leading to the curve shown in Figure 5.
S
05 10 15 2025 30 35 40
2
3
4
5
6
7
8
9
10
11
12 x 10
5
Channel indice
Variance
Figure 5. The variance evolution of the hyperspectral se-
quence.
From this variance evolution, one can point out an
important increasing tendency along the channels. Mor-
eover, one can sort the channels so that those presenting
low values are potentially inte rested to be mixed with the
remaining channels in th e context of Multimodal Co mpr-
ession.
3.2. Multimodal Compression Based 3D-SPIHT
Phase
By setting the filling area to 20%, producing hence a ROI
of 80%; in such a case, four images can be inserted in the
remaining images (initially 36) of hyperspectral sequ-
ences. After this reduction, a sequence of 32 images is
obtained. This new sequence has been compressed using
3D-SPIHT within a range [0.01 - 1.75] bpppb. For each
bit-rate, the averaged , denoted by A, the
root mean squared error (RMSE) and percentage error
(%E) are evaluated [13]. They are given by the Equations
(7) and (8), respectively:
PSNR PSNR
 
36614 5122
,,
11 1
1
36 614 512ij ij
ki j
RMSEIkR k
 








(7)
36614 512,,
11 1,
() ()
1
%36614512( )
ij ij
kij ij
I
kRk
EIk
 





 (8)
where


,ij
I
k and


,ij
Rkare the pixel values the origi-
nal and the reconstructed base respectively at the spatial
location


,ijof the band k.
36: number of sequences to be compressed.
512 × 614: size of each image from the dataset of the
Yellowstone scene.
Table 1 lists the RMSE and %E of the reconstructed
data for different bpppb.
In Figure 6, only the A evaluated on the whole
sequence is shown. Therefore, three performance curves
are provided, corresponding, respectively to 1) the origi-
nal sequence compressed by 3D-SPIHT; 2) reduced se-
quence using the Multimodal Compression; 3) reduced
sequence using the Multimodal Compression for which
the A is evaluated only on ROIs. By analyzing
these curves, based only on the A, one should note
that for low bpps, the quality of the decompressed im-
ages is objectively almost the same, whereas for high
bpppbs, the quality remains subj ectively the same (based
on the visual quality). This can be explained by the fact
that when dealing with PSNRs greater than 50 db (which
is the case here), it becomes very difficult to distinguish,
visually between image qualities.
PSNR
PSNR PSNR
On the other hand, when comparing the performance
curve corresponding to the direct compression (3D-SPI-
HT) to the MC-3D-SPIHT by considering the A at
the level of ROIs, one can notice that almost the same
PSNR

Fast Encoding-Decoding of 3D Hyperspectral Images Using a Non-Supervised Multimodal Compression Scheme 320
results are obtained. By considering the other criteria of
quality evaluatio n, namely the RMSE and E%, the concl-
usion is the same as it is highlighted in Tabl e 1. Very
close values are obtained.
In terms of quality, one can conclude that the MC-3D-
SPIHT preserve the information without any significant
distortion (see Figure 7). In terms of computing time, the
proposed technique becomes particularly interesting
since the encoding or the decoding are both achieved on
a reduced number of images. As shown in Table 2, en-
coding/decoding time is compared for various bpppbs.
The evaluation has been performed on a computer run-
ning at 1.6 GHz. Table 2 highlights the computing time
for both techniques and show clearly that the MC-3D-
SPIHT outperforms the direct compression technique
which is even obvious since only 32 images are consid-
ered. From these 32 images, 4 extra images are extracted
by the proposed appro ach without significant lo ss of qua-
lity.
On the other hand, the mixing/separation functions are
not time consuming since these tasks can be achieved
using a DMA “Direct Memory Access” which doesn’t
Table 1. Statistical based measures of reconstructed data
for different bpppb (with and without multimodal com-
pression).
RMSE Percentage Error (%)
bpppb 3D-
SPIHT MC-3DS PIHT MC-3DSPIH T (ROI) 3D- SPIHT MC-3D SPIHT MC-3DSPIH T (ROI)
0.01 18.99 22.60 18.48 16.65 30.93 16.99
0.1 10.53 15.93 13.33 5.50 13.66 7.52
0.5 6.71 10.84 7.57 2.65 5.16 2.71
1 5.24 9.86 5.57 1.58 3.34 1.51
00.20.4 0.60.8 11.2 1.41.6 1.82
40
45
50
55
60
65
70
75
Bit-rate(bpppb)
PSNR
A
(dB)
Ori ginal se quen ce
Reduced s equenc e
Reduced s equenc e
(P S NR
A
is evaluat ed only on ROIs)
Figure 6. Comparison between the direct compression using
3D-SPHIT, MC-3D-SPIHT. In low bpppbs the quality is
objectively almost the same. In high bpppbs the visual qual-
ity is also almost the same (>55 dB). Moreover, MC-3D-
SPIHT is faster in the encoding/decoding process.
(a) (b)
(c) (d)
Figure 7. Reconstitution of image channel 30. (a) Original
image; (b) Same image embedding other pixels from an-
other image (in the contours); (c) Reconstructed image at
0.5 bpp; (d) Reconstructed image at 1 bpp.
Table 2. Comparison of the encoding and decoding times
evaluated for different bpppb (with and without multimo-
dal compression).
Without M-Compression With M-Compression
bpppb Tc T
dec T c T dec
0.01 9.55 3.16 9.45 3.15
0.1 10.221 3.36 10.08 3.25
0.25 10.31 3.70 10.38 3.55
0.5 10.98 4.22 10.95 4.01
1 12.61 5.30 12.09 5.05
1.5 14.17 6.80 13.53 6.28
2 15.67 7.61 15.08 7.39
Tc: encoding times on seconds. Tdec: decoding times on seconds.
require to use any processor cycle. Finally, the proposed
technique requires an interpolation to be applied on each
image contour, but the time required to achieve this task
is insignificant (<500 Milliseconds) compared to the
whole decoding time process.
Objectively, using the multimodal compression one
can encode a given volume by preserving the same qu al-
ity as obtained with the direct techniques but using, es-
sentially, less computing time which is an important ad-
vantage.
4. Conclusions
In this work we presented a new technique to compress
hyperspectral image sequences using a Multimodal Co m-
pression approach. The proposed scheme inclu des a mix-
ing function, a separation function and the 3D-SPIHT al-
gorithm. Tests have been performed in two main phases.
In the first phase, we have shown that the 3D-SPIHT (w-
hich considers hyperspectral sequences as a volumetric
image) outper forms bo th the JPEG 2000 and CCSD S sta-

Fast Encoding-Decoding of 3D Hyperspectral Images Using a Non-Supervised Multimodal Compression Scheme
321
ndards. In the second phase, the size of the sequence to
be compressed has been reduced by approximately 20%
using the mixing function then compressed using 3D-
SPIHT. After the decompression and separation process,
we showed that the quality of the decoded images using
various criteria are objectively and subjectively very
close to the one obtained by a direct compression. The
major advantage is that the MC-3D-SPIHT reduces sig-
nificantly the coding/decoding time and improves the
compression ratio in comparison to a direct compression
using JPEG 2000 or CCSDS. This makes this approach
very appropriate to deal with huge data. In the future
work, optimizing the mixing function could be an inter-
esting perspect ive .
REFERENCES
[1] E. Christophe, C. Mailhes and P. Duhamel, “Hyperspec-
tral Image Compression: Adapting SPIHT and EZW to
Anisotropic 3D Wavelet Coding,” IEEE Transactions on
Image Processing, Vol. 17, No. 12, 2008, pp. 2334-2346.
doi:10.1109/TIP.2008.2005824
[2] B. Penna, T. Tillo, E. Magli and G. Olmo, “Transform
Coding Techniques for Lossy Hyperspectral Data Com-
pression,” IEEE Transactions on Geoscience and Remote
Sensing, Vol. 45, No. 5, 2007, pp. 1408-1421.
[3] M. Lahdir, S. Ameur and A. Adane, “Algorithme non ité-
ratif basés sur les ondelettes biorthogonales et les fract-
ales pour la compression des images satellitaires,” Té-
lédétection, Vol. 6, No 4, 2006, pp. 345-360.
[4] D. S. Taubman and M. W. Marcellin, “JPEG2000: Image
Compression Fundamentals, Standards and Practice,” Klu-
we r Academic Publishers, Boston, 2002.
[5] Q. Du and J. E. Fowler, “Hyperspectral Image Compres-
sion Using JPEG2000 and Principal Component Analy-
sis,” IEEE Geoscience and Remote Sensing Letters, Vol.
4, No. 2, 2007, pp. 201-205.
doi:10.1109/LGRS.2006.888109
[6] P. L. Dragotti, P. Giovanni and A. R. P. Ragozini, “Com-
pression of Multispectral Images by Three Dimensional
SPIHT Algorithm,” IEEE Transactions on Geoscience and
Remote Sensing, Vol. 38, No. 1, 2000, pp. 416-428.
doi:10.1109/36.823937
[7] J. E. Fowler and J. T. Rucker, “3D Wavelet-Based Com-
pression of Hyperspectral Imagery,” In: C.-I. Chang, Ed.,
Hyperspectral Data Exploitation: Theory and Applica-
tions, John Wiley & Sons, Inc., Hoboken, 2007.
[8] A. Naït-Ali and C. Cavaro-Menard (Ed.), “Compression
of Biomedical Images and Signals,” ISTE-John Wiley
and Sons, London, 2008, pp. 247-275.
[9] A. Naït-Ali, E. H. Zeybek and X. Drouot, “Introduction to
Multimodal Compression of Biomedical Data,” In: A.
Naït-Ali, Ed., Advanced Biosignal Processing, Springer,
Berlin, 2009, pp. 353-375.
[10] X. Tang and W. A. Pearlman, “Three-Dimensional Wave-
let-Based Compression of Hyperspectral Images,” Chap-
ter in Hyperspectral Data Compression, Kluwer Aca-
demic Publishers, Boston, 2005.
http://www.cipr.rpi.edu/ pearlman
[11] A. Said and W. A. Pearlman, “A New Fast and Efficient
Image Codec Based on Set Partitioning in Hierarchical
Trees,” IEEE Transactions on Circuits and Systems for
Video Technology, Vol. 6, No. 3, 1996, pp. 243-250.
doi:10.1109/76.499834
[12] P. S. Yeh, G. Moury and P. Armbruster, “The CCSDS
Data Compression Recommendation: Development and
Status,” Proceedings of SPIE Application of Digital Im-
age Processing, Seattle, 7-10 July 2002.
[13] S.-E. Qian, J. Lévesque and R. A. Neville, “Evaluation of
Noise Removal of Radiance Data Onboard Data Com-
pression of Hyperspetral Imagery,” WSEAS International
Conference on Remote Sensing, Venice, 2-4 November
2005, pp. 37-42.