Architectural Model for Visualization of High Definition Images on
Mobile Devices
Germán Corredor
1
, Daniel Martínez
1
, Eduardo Romero
1
and Marcela Iregui
2*
1
Universidad Nacional de Colombia, Carrera 30 # 45-03, Bogotá D.C., Colombia
2
Universidad Militar Nueva Granada, Carrera 11 # 101-80, Bogotá D.C., Colombia
Keywords: Architecture, Decoding, Images, Interaction, JPEG2000, Mobile Devices, Protocol, Multimedia, Streaming,
Visualization.
Abstract: In recent years, the mobile device demand has largely increased because of the accessibility, ubiquity and
portability of such devices, which are being used not only for personal purposes but also in several
applications like education, science, entertainment, commerce and industry, among others. Visualization
and interaction with high definition multimedia content, like large images and videos, using mobile devices,
represents a challenge because of their very limited machine resources and bandwidth. For such application,
this content requires special treatment so that users can properly access and interact. In this article, it is
proposed an architectural model for efficient streaming and visualization of very large images on mobile
devices using the JPEG2000 standard and an adapted image transfer protocol. Results show that the
introduced architecture is effective for visualizing regions of large images and presents good performance,
both for transmission and decoding processes, allowing a simple and dynamic interaction between user and
images.
1 INTRODUCTION
Real-time access to information has become an
important task in different daily activities; this is
why the use of portable devices and mobile services
has increased in recent years. Portability and last
hardware advances of mobile devices have induced
people to use several mobile applications for
managing personal information, remote information
access, multimedia services, etc. (Buchinger et al.,
2011). Furthermore, currently, the importance of
mobile devices is beyond the personal use, because
they are starting to be extensively used in fields like
education, science and research, entertainment,
commerce, industry, among others (Qiao et al.,
2008).
The increasing demand of mobile devices has
promoted development of more sophisticated
devices, with larger capabilities, which have broaden
the scope and possibilities. In despite of the latter
technological advances in such devices, they still
have several external limitations such as
communication networks, geographic access and
internal limitations such as memory, CPU power
*Corresponding author
and display size (Agu et al., 2005). All those
limitations make difficult to load, decode and
visualize multimedia content such as large images
and videos. In some fields, like microscopy or
cartography, it is necessary to work with high
definition images, characterized by its large size,
high resolution and quality. These images are
represented by a very large volume of data, so, for
visualizing them on mobile devices it is necessary to
give them a special treatment to allow the users to
properly access and interact.
To ease large image streaming and visualization,
different transmission and coding algorithms have
been developed, which facilitate storage,
transmission and display of graphic content. In
particular, the JPEG standard has been used for a
long for coding images; however, current necessities
require using more flexible and efficient standards
(Sarraf and Wakim, 2007). In 2000, the Joint
Photographic Expert Group (JPEG) published the
JPEG2000 (J2K) standard, which, based on the
concept of regions of interest and scalability,
represents advances for the image compression
technology, for which the image coding system has
been optimized not only for being efficient, but also
163
Corredor G., Martínez D., Romero E. and Iregui M..
Architectural Model for Visualization of High Definition Images on Mobile Devices.
DOI: 10.5220/0004052601630170
In Proceedings of the International Conference on Signal Processing and Multimedia Applications and Wireless Information Networks and Systems
(SIGMAP-2012), pages 163-170
ISBN: 978-989-8565-25-9
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
for being scalable and interoperable in network and
mobile environments (Sarraf and Wakim, 2007;
Rosenbaum and H. Schumann, 2006). Levels of
detail, regions of interest and progressive
transmission are popular concepts for handling
graphical data in resource-limited environments
(Rauschenbach and Schumann, 1999). Thererfore,
the J2K standard is presented as a suitable tool to
address a proper streaming and visualization on
mobile devices.
Several works have explored the possibilities to
access to J2K content by optimizing the interactive
navigation, by proposing distribution protocols,
management of client/server platforms, cache and
prefetching strategies, among others (Taubman and
Rosenbaum, 2003; Descampe, et al., 2007); (Iregui
et al., 2007); (Deshpande and Zeng, 2001); (Iregui et
al., 2002); (Iregui et al., 2002); (Meessen et al.,
2003); (Moshfeghi and Ta, 2004). However these
works do not consider the problem of minimal
available capabilities of mobile devices.
In the mobile field, some works have focused on
areas such as reliability and error resilience over
noisy channels (Ho and Kahn, 1997) and content
delivery security (Díaz et al., 2006). Liu et al (2003)
presented a model for browsing of images on small
displays, however they work with images of range of
1600x1200 pixels; this model is not applicable to
very large images, as satellite imaging, which may
have sizes of more than 15000x15000 pixels.
Rosenbaum and Schumann (2006) proposed a model
for viewer guidance for mobile devices by exploiting
the J2K features; their solution requires to filling the
omitted code-stream positions with predefined data
to keep it compliant and can be decoded, however,
this solution is not applicable to very large images
because if the area to be processed and decoded is
very large, the required processing time and memory
may exceed the capacities of the device. Otherwise,
Google Earth is a well-known application for
accessing to large satellite images that runs on
desktop computers and mobile devices. It uses a
very large set of hierarchy prerendered tiles of
different spatial resolutions, then, each time a user
increases the level detail of the image, a new set of
images is loaded (MicroImages, Inc., 2010). In this
way, the application requires a complex organization
and hierarchy of files and directories and a big
amount of space to store the whole image.
Furthermore, this application only works with the
provided images and does not allow using different
ones.
The present investigation introduces
anarchitectural model to allow optimal interaction
with very large images of general purpose from
mobile devices. The proposed model facilitates a
mobile client to browse very large remote images by
displaying regions in a flexible and scalable way,
offering a nearly seamless navigation, adapted to the
restricted capabilities of the devices and the channel
bandwidth. This represents an advantage over
available applications that do not allow accessing
and navigation in very large images or regions in full
quality and resolution. Moreover, the proposed
architecture is modular and decoder independent,
permitting thereby easy adaptation to new models
and specialized applications.
This paper is organized as follows. In section 2, a
brief overview of JPEG2000 standard is presented.
Section 3 introduces the proposed model for
streaming and visualization of high definition
images on mobile devices. In Section 4, an
implementation of the proposed model is presented.
In section 5, experimental results are reported,
providing evidence of the performance and efficacy
of the model. Finally, section 6 presents brief
conclusions.
2 JPEG2000 OVERVIEW
JPEG2000 is an image compression standard
designed by the Joint Photographic Expert Group,
based on the Discrete Wavelet Transform and the
EBCOT encoder (ISO/IEC 15444-1, 2000). This
standard provides several advantages such as
improved compression efficiency, lossy and lossless
compression, multiple resolution representation,
random code-stream access and processing and
quality refinement (Rabbani and Rajan, 2002).
A single J2K data stream typically contains
numerous embedded subsets, which may be
extracted to recover a portion of the original image
at any of a large number of different spatial
resolutions, image quality layers, or in selected
spatial regions. A J2K image is split into rectangular
regions that are encoded independently, called tiles,
but also defines collections of spatially adjacent
code-blocks, known as precincts. Each precinct is
represented as a collection of packets, with one
packet for each quality layer, resolution level and
component. These embedded compressed data
subsets allow a low quality or low resolution image,
or one whose details cover only a small spatial
region, to be incrementally improved by adding the
missing elements from the compressed data stream
(Taubman, 2002).
SIGMAP2012-InternationalConferenceonSignalProcessingandMultimediaApplications
164
3 SYSTEM ARCHITECTURE
The proposed application aims to be as simple as
possible in order to reach acceptable response times,
provided the highly limited memory, processing
power and low bandwidth of mobile devices, among
other critical constraints. The main idea is to allow
the user to navigate in a seamless way by requesting
partial content of an image, according to each region
of interest, instead of processing the whole image; in
such scenario, only data contributing to the
requested representation are transmitted and
decoded.
Through the exploitation of the granularity and
scalability of J2K data streams, the most appropriate
subsets are delivered from a server to an interactive
client, incrementally improving the image quality
and/or resolution in a consistent manner with the
client’s interests at any given time (Taubman, 2002).
As mentioned in the previous section, J2K
codestream is conformed by packets, one for each
quality layer, resolution level, component and
position. In this way, only those packets belonging
to a region at a specific resolution level and quality
percent of the image could be retrieved. Because
tiles compress less efficiently, introduce unpleasant
boundary artifacts at low bit-rates and provide poor
resolution scaling properties (Taubman, 2002) and
because of packets flexibility, in this work, packets
are adopted as the fundamental unit of data
exchange between server and client.
For this purpose, the server, which stores the set
of still encoded images, must be capable of
extracting packets from images according to a given
request. To enable random access to specific
portions of the codestream in a fast way an index file
for each image has been off-line generated based on
the JPIP specification (ISO/IEC 15444-9, 2000),
facilitating a minimal memory use and reduced disk
access. The index is a simple text file containing the
information organized into two boxes: image
information box and packet information box. The
former contains information such as width and
height, progression order id, number of components,
number of quality layers, number of decomposition
levels, etc. The latter contains the index, the features
(i.e. tile, layer, resolution, component and precinct),
and the starting and ending positions of each image
packet. The server is also responsible for generating
a sequence of J2K packets and delivering it to the
client, which is located on a mobile device.
The client has a Graphic User Interface (GUI)
that allows the user to facilitate the comprehension
of the content through an appropriate representation
and means for interaction, by specifying which
regions of a particular image must be displayed with
a certain degree of detail. The GUI enables to set up
the Window of Interest (WoI) coordinates, the
resolution level and the quality percent. For this
purpose, a method that calculates the necessary
packets to meet the user requirement was designed.
Since most of the data may be reused and a non-
redundant data transmission is desired, a cache
management is performed by storing the received
packets; in this way, only missing data are requested
to the server.
Although the J2K compression algorithm and its
packet structure provide excellent tools on which to
base an interactive image application, the data
syntax described in the standard excludes the
interactive construction of a valid data stream from
arbitrarily ordered packets (Taubman, 2002). As not
every available image data might be needed, it is
necessary to do some modifications to data to keep
compliance; previous works have addressed this
issue by filling the omitted code-stream positions
with predefined data (Rosenbaum and Schumann,
2006). However, in this work it is proposed a
different strategy: based on the information of the
requested region, the image main header is modified
as to include specifically the packets required to
reconstruct the queried ROI. In other words, the
image size, quality layers and resolution levels are
set to the ROI. In this way, the decoding process is
faster and efficient because the transcoding process
requires only little computing power and memory,
and because the decoder do not have to process trash
or empty data. This strategy allows using the
decoder with no change, to decompress a set of
packets; so that any J2K decoder can be used.
3.1 Server
The server stores the J2K images and their
respective index files and thumbnails. It sends
packets and information of the images to the client
through a TCP/IP channel.
User interface (UI): It is an interface through
which the system’s administrator can manage the
server. It naturally allows the addition of new images
to make them available to the client.
J2K Encoder: It is responsible for encoding the
images in the J2K standard. It uses a predefined set
of parameters to perform encoding. Once the image
is encoded, it is stored as a file at a specific location
on the server.
Index generator: It generates an index file for
each codified image. The index file is used to get
ArchitecturalModelforVisualizationofHighDefinitionImagesonMobileDevices
165
Figure 1: Top-level runtime view of the architecture.
image information and to ease random access to
specific portions of the codestream.
Thumb generator: It generates a thumbnail file
for each image. The thumbnails are used to rapidly
allow the client to access and view the available
images.
Server connector: It is responsible for
communication with the client. It receives and
processes the client requests and sends packets and
image data.
3.2 Client
The client is located in a mobile device that allows
the user to interact through a GUI by requesting and
viewing regions of interest at certain level of
resolution and quality.
Client Connector: It is responsible for
communication with the server. It sends request
messages to the server and sends the received
response data to the RP.
Request Processor (RP): The RP is an interface
between the GUI and the Client Connector. It is
responsible for processing the requests from the
GUI. It sends the corresponding request to the Client
Connector, sends and retrieves packets from the
Packet Cache Manager, generates a compliant J2K
codestream to send it to the J2K decoder and sends
the generated pixels to the GUI.
Packet Cache manager: It stores all the received
J2K packets and recover them when are requested
by the RP. The cache size is parameterizable and
may be set depending of the capacity of the device.
The packet cache manager currently uses a Least
Recently Used policy to discard packets and free
memory when it is almost full.
J2K Decoder: The J2K decoder is responsible for
decoding the requested image region. It receives as
input a transcoded compliant bitstream and returns
the set of pixels corresponding to the image
representation.
Pixel Cache manager: It stores the decoded
pixels. It currently uses a policy to discard packets
based on the distance; it stores the adjacent areas of
a user request and deletes them when the user is
centred in a non-adjacent (distant) region when
memory is almost full. The size of the pixel cache is
also parameterizable.
Graphic User Interface (GUI): It allows the user
to specify which regions of an image should be
transmitted, at which degree of detail.
4 SYSTEM FOR STREAMING
AND VISUALIZATION OF
IMAGES
A simple prototype was developed in order to
validate the proposed architecture. Hereafter, there is
a brief description of the features of the developed
application. Since the architecture consists of a
client and a server, we developed a stand-alone
application for each node.
4.1 Server Implementation
As mentioned in the previous section, the server
stores the J2K images and their respective index
files and thumbnails, and allows communication
with the client. The server was developed in the Java
platform, Standard Edition, version 1.6.
Uncompressed images and coding parameters
are sent to the J2K encoder, which generates a coded
image file; encoding was performed with the Jasper
J2K implementation (Adams and Kossentini, 2007).
The start of packet (SOP) marker was used to
facilitate the index file generation. The index
generator receives as input an encoded image; it
reads the markers of the image and extracts the
necessary information to generate the index file.
Finally, the Server Connector is provided with a
socket that receives and processes the client requests
and sends the corresponding responses.
SIGMAP2012-InternationalConferenceonSignalProcessingandMultimediaApplications
166
4.2 Client Implementation
The client was developed in the Android platform,
version 2.2. The client runs a Graphic User
Interface, a Request Processor, a J2K decoder, a
Pixel Cache Manager, a Packet Cache Manager and
a Client Connector. The communication with the
server is done through a socket that the Client
Connector manages.
The GUI allows the client to connect to the
server by typing the corresponding IP address. Once
the connection is established, it is requested the list
of available images with their respective thumbnail,
which are shown to the user through the GUI. When
the user selects one image, it is requested the load of
that one and the server sends its information (size,
number of resolutions, quality layers, etc.) and its
main header. Next, the user selects a region of
interest by setting the coordinates, a level of
resolution and quality percent. If the requested
region is not in the pixel cache, the request is sent to
the RP, which calculates the necessary packets to
meet that requirement and looks at which packets
are stored in packet cache memory, in this way, it
only requests to the server the missing data. Once
the client has the necessary packets, either from the
server or from the cache, they are transcoded into a
compliant codestream. As mentioned in the previous
section, to let the packets to be decoded, it is
necessary to modify the image main header. For the
decoding process it was used the Jasper Software, so
it was developed an interface function that allows to
send to Jasper the codestream and to receive the
corresponding pixels. Finally, the pixels are sent to
the graphic interface to be presented to the user, that
pixels are stored on the pixel cache.
5 EXPERIMENTAL RESULTS
Two nodes were used to perform the experiments: A
server and a mobile device. The former was a
desktop computer with operating system Windows
7, 4 GB RAM memory and 2.71 GHz dual-core
processor. The latter was the Amazon Kindle Fire, a
tablet with operating system Android 2.3, 1024×600
display size, 512 MB RAM memory and 1 GHz dual
core processor. The mobile device was connected to
a Wi-Fi network with a bandwidth of 25 Kbps.
The experiments were performed with an
uncompressed satellite image of 786433 KB and
resolution 16384x16384 pixels. The image was J2K
compressed by using the following coding
parameters. LRCP progression order, 3 components,
10 quality layers, 4 resolution levels (5 resolutions),
lossless codification, precinct size of 64x64 for each
resolution and codeblock size of 64x64. The final
size of that image was 125544 KB, reaching a
compression rate of 84.04%. The generated index
for this image has a size of 99629 KB and the
reading and parsing process is about 7 seconds.
Because the index is stored in the server side, its
format was not optimized for storage size.
5.1 Comparing JPEG and JPEG2000
In the first experiment, the performance of streaming
and visualization by using JPEG and JPEG2000
standards was compared. As the JPEG standard does
not provide multiresolution representation, a JPEG
pyramid was constructed, i.e., the original image
was coded with 5 different image sizes, each
corresponding to a different resolution level. The
total size of the 5 JPEG images was 170529 KB,
reaching a compression of 78.32%.
Transmission and processing time were
compared for a set of user requests. In this case, the
user starts at a certain region of interest and zoom in
over it, from the first resolution to the fourth, at full
quality. Results are shown in Table 1.
Results show that the number of transmitted
bytes for a JPEG image is greater than the
JPEG2000 image; due the JPEG standard does not
allow a straight access to regions of the image, it is
necessary to request and to decode the whole image
in order to present it or a region to the user.
This represents an issue when the transmitted
image is too large, because it is required a long time
for downloading it and because the device may not
have enough memory to store and process it.
Furthermore, processing time for JPEG is less than
for JPEG2000 for small size images; however, for
large images in the JPEG format, due the limitation
in memory and processing power, the application
generated an error and could not decode and show
the image representation.
5.2 Simulating a User Browsing
Protocol
For this experiment, memory usage and transmission
and computing times were measured for a set of user
requests by simulating a browsing protocol for
which a user scrolls over adjacent and overlapping
regions. The user starts viewing an image in the
lowest resolution level and 50% of quality. Next, the
user finds an interest area, zooms in until the fourth
resolution level. Finally, as the user stops at this
ArchitecturalModelforVisualizationofHighDefinitionImagesonMobileDevices
167
Table 1: Bytes and decoding and computing times for a satellite image by using the JPEG and the JPEG2000 standards.
N/D (No data) means that the request was not processed because of memory overflow.
Resolution
Level
JPEG JPEG2000
Transm.
bytes
Transm. time
(ms)
Computing time
(ms)
Transm.
bytes
Transm. time
(ms)
Computing time
(ms)
1 576288 710,4±134,42 147,1±36,82 283472 2102,6±622,36 1660±158,73
2 2334063 2332±353,16 469,7±11,21 210359 1466,8±141,65 1602±15,82
3 9310361 8355,3±1582,5 N/D 216217 1808,5±764,87 1747,3±285,53
4 35151251 N/D N/D 199148 1446,1±21,87 1582,7±15,83
area, a quality refinement is evaluated by requesting
the remaining quality layers, until the maximum
quality level. The browsing protocol is shown in
figure 2.
Figure 2: Representation of the requested regions. Dashed
areas correspond to regions to be requested, while shaded
áreas correspond to previously requested areas, stored in
cache. Figure (a) shows the first request in which there are
not data stored in cache. Figure (b) shows the request of an
adjacent region to the first one. Figures (c) to (e) shows
requests of areas that overlap previously requested areas.
Figures (f) to (h) shows requests of areas simulating a
zoom in, since the second resolution level to the fourth
one. Figures (i) to (m) shows requests for a quality
refinement.
While the original uncompressed image is 768
MB in size, at the end of the browsing protocol, only
952 KB of compressed data are transmitted.
Transmission of the packets took only 4s and the
image processing took only 14s for the whole
navigation.
5.2.1 Measurement of Transmission and
Computing times
In this experiment, the transmission and computing
times were measured for the browsing protocol.
Figure 3 presents the results of this experiment and
includes the accumulated time of transmission and
computing.
Requests (a) to (e) represent the scrolling over
adjacent and overlapping areas. In this case, it was
used a packet cache policy because some regions
intersect with previously requested regions. Once the
graphic area representation is obtained, it is added to
the screen to compose the whole region. Response
times of the first request are relatively high because
there are not data in cache. As the area of the second
region does not overlap with the first, every packet
has to be transmitted and decoded. For next requests,
transmission and decoding times are lower because
of the intersected regions.
Figure 3: Variation of transmission and decoding times for
a browsing protocol. Requests (c) to (e) represents regions
that overlap with previously requested data, so, less
amount of data must be transmitted and processed.
Requests (f) to (h) represents a zoom in over a region.
Requests (i) to (m) represents a quality refinement
process.
Requests (f) to (h) represent a zoom in operation
over a specific area. The requested regions
correspond to areas of 512x512 pixels. Results show
that the transmission time slightly increases because
the use of packet cache. Regarding the decoding
process, given that the requested area is lower than
the device’s display size, it is responding in proper
times. Other experiments have shown that requesting
areas larger than the display may have repercussions
on the system performance. It is highly
recommended that requested areas do not double the
screen size.
Request (i) to (m) stand for a quality refinement
operation over a specific area. Here, the user was
starting at a specific region with 50% of quality, and
SIGMAP2012-InternationalConferenceonSignalProcessingandMultimediaApplications
168
progressively the system requests and decodes the
layers corresponding to higher quality percentages.
On the one hand, this process takes advantage of the
cache policy by using the previously requested
packets so that the transmission times are reduced.
On the other hand, decoding time increases for a
quality layer with respect to the previous. This
happens when decoding a new quality because it
requires using the whole data of the previous layer
and there is not a direct method to perform cache
management for quality layers. Results show that
response times increases when the quality
percentage increases, however, the system offers the
user flexibility to select the desired maximum
quality for each request, according to his/her
necessities; in this way, better response times can be
obtained.
5.2.2 Memory Usage Evolution
In this experiment was measured the memory usage
of the application for the browsing protocol
previously depicted. The memory was measured
each second during the execution of the browsing
session, for a total of 20 seconds. Results are shown
in figure 4.
Figure 4: Evolution of memory usage of the application
for a browsing protocol.
Results show that the application memory usage
keeps between about 15 and 25 MB, which
represents only 4.88% of the total memory of the
device. That means, the application is using only a
small part of the device’s memory, so, the memory
usage is efficient and it is applicable to most of the
current devices.
General results show that the introduced
architecture not only allows visualizing regions of
very large images but also presents a good
performance, both for transmission and decoding
processes. The procedure to determine every packet
contributing to a desired ROI requires low
computing power and decoding time is reduced
because the area of the requested image is not larger
than the display size. It was defined a maximum
response time to visualize a requested area of 5
seconds and the results shown the user do not have
to wait more than 2 seconds. Additionally, a
software-based decoder was used, however, if a
hardware-based one is used, response times may
improve in a meaningful way. The visualization is
adaptable to user necessities; because it can display
the best quality and the best resolution level adapted
to the possibilities of the device (memory,
computing power and screen size). Also, it was
demonstrated that use of J2K is most appropriated
for large images on mobile devices than JPEG. The
use of J2K has important benefits; the standard
inherently provides a high compression
performance, but also a structure of the encoded data
which allows accessing the stream only in those
packets contributing to the required ROI and thereby
much bandwidth can be saved because it is not
necessary to transmit and process the whole image.
In the server side, the use of J2K standard gives an
improved compression efficiency, which allows
storing several large images and their respective
indexes without requiring a lot of disk space.
6 CONCLUSIONS
In this paper, an architecture model for streaming
and visualization of high definition images,
characterized by its high resolution, quality and size,
on mobile devices was presented. The proposed
model allows a user to request regions of the image
at certain level of resolution and quality. The
proposed solution takes advantage of the J2K
granularity to allow accessing to high definition
images to make an optimal content delivery to
mobile devices with different capabilities and
bandwidth. The client only decodes the required data
to generate the representation of the image region, so
the memory and processing usage is minimized and
the system presents proper response times. Due to
the modular design and the loosely coupled decoder,
it is easy to adapt the architecture to new models and
to develop specialized applications. Future work
includes index generation optimization and
implementing cache strategies based on user
navigation.
REFERENCES
ISO/IEC 15444-1, 2000. Information technology -
JPEG2000 image coding system – Part 1: Core
ArchitecturalModelforVisualizationofHighDefinitionImagesonMobileDevices
169
coding system.
ISO/IEC 15444-9, 2004. Information technology -
JPEG2000 image coding system – Part 9: Interactivity
tools, APIs and protocols
Rabbani and Rajan, 2002. An overview of the JPEG2000
still image compression standard. Signal Processing:
Image communication, 17:3–48.
Taubman and Rosenbaum, 2003. Rate distortion optimized
interactive browsing of JPEG2000 images. In IEEE
International Conference on Image Processing
(ICIP2003), volume 3, pages 765–768.
Descampe, et al., 2007. Pre-fetching and caching
strategies for remote and interactive browsing of
JPEG2000 images. IEEE Trans. on Image Processing,
16(5):1339-1354.
Iregui, Gómez and Romero, 2007. Strategies for efficient
virtual microscopy in pathological samples using
JPEG2000, Micron Vol 38 pag 700-713.
Liu et al, 2003. Automatic browsing of large pictures on
mobile devices. Proceedings of the 11th ACM
international conference on Multimedia.
Qiao, Feng and Zhou, 2008. Information presentation on
mobile devices: Techniques and practices. Lecture
Notes in Computer Science, Volume 4976/2008, 395-
406.
Burigat, Chittaro and Gabrielli, 2008. Navigation
techniques for small-screen devices: An evaluation on
maps and web pages. International Journal of Human-
Computer Studies, Vol. 66, No. 2, pp. 78-97.
Agu et al., 2005. A Middleware architecture for mobile 3D
graphics. Proceedings of the 25th IEEE International
Conference on Distributed Computing Systems
Workshops.
Buchinger et al, 2011. A survey on user studies and
technical aspects of mobile multimedia applications,
Entertainm. Comput., doi:10.1016/j.entcom.
2011.02.001
Rauschenbach and Schumann, 1999. Demand-driven
image transmission with levels of detail and regions of
interest. Computers & Graphics 23 857-866.
Sarraf and Wakim, 2007. Improving JPEG2000 images
delivery over GPRS mobile networks. Proceedings of
the 6th WSEAS Int. Conf. on Electronics, Hardware,
Wireless and Optical Communications, Corfu Island,
Greece.
Rosenbaum and Schumann, 2006. JPEG2000 based
viewer guidance for mobile image browsing. 12th
International Multi-Media Modelling Conference
Proceedings.
Taubman, 2002. Remote browsing of JPEG2000 images.
Proceedings IEEE ICIP.
Deshpande and Zeng, 2001. Scalable streaming of
JPEG2000 images using Hypertext Transfer Protocol.
Proceedings ACM Multimedia.
Iregui, Chevalier and Macq, 2002. Optimal caching
mechanisms for JPEG 2000 communications.
EUSIPCO 2002 - European Signal Processing
Conference, Toulouse, France, Sept. 2002,
Proceedings Vol. 3, pp. 201-204
Iregui et al., 2002. Flexible access to JPEG2000
codestreams.
23rd Symposium on Information Theory
in the Benelux, May 29-31, 2002, Louvain-la-Neuve,
Belgium
Meessen et al, 2003. Layered architecture for navigation in
JPEG2000 Mega-Images. WIAMIS 2003 - 4th
European Workshop in Image Analysis for Multimedia
Interactive Services, April 9-11, London, UK, Proc.,
pp. 92-95
Moshfeghi and Ta, 2004. Eficient image browsing with
jpeg2000 internet protocol. In SPIE: Medical Imaging
2004: PACS and Imaging Informatics.
Díaz et al., 2006. JPEG2000 images protection sent to
mobile devices. Proceedings of the I International
Conference on Ubiquitous Computing: Applications,
Technology and Social Issues. Spain.
Ho and Kahn, 1997. Image transmission over noisy
channels using multicarrier modulation. Signal
Processing: Image Communication, Volume 9,
Number 2, January 1997 , pp. 159-169
Hasan and Kim, 2009. An automatic image browsing
technique for small display users. 11th International
Conference on Advanced Communication Technology.
Adams and Kossentini, 2007. Jasper: A software-based
jpeg-2000 codec implementation. Available at
<http://www.ece.uvic.ca/~frodo/jasper/>
MicroImages, Inc., 2010. Google Earth Structure.
Available at <http://www.microimages.com/docu
mentation/TechGuides/76googleEarthStruc.pdf>
SIGMAP2012-InternationalConferenceonSignalProcessingandMultimediaApplications
170