07-04-2012, 11:45 AM
An Efficient Image Processing Method Based on Web
Services for Mobile Devices
An Efficient Image Processing Method Based on Web.pdf (Size: 275.97 KB / Downloads: 154)
Abstract
the resource limitation of mobile devices causes the
problem that the existing image processing software based on the
centralized computing mode had difficulty running in mobile
devices. A solution is given in the paper by adopting web servicebased
image processing method. For one thing, image processing
tasks were distributed to service providers’ service registry and
service requesters. For another, what the service providers
should do was only to invocate the specific image processing
services provided by service providers. Consequently, web
service-based solution reduces the resource consumption of
mobile devices by redistributing image processing tasks.
Compared with traditional methods of image processing, Web
service-based image processing method has the advantages of
loose coupling and component oriented and can take full
advantage of the computing resources in heterogeneous network.
Thus web service-based image processing method can effectively
solve the resource bottle-neck that traditional image processing
software had.
Keywords-Web Service; image process; Qos; UDDI; mobile
devices;
I. INTRODUCTION
The advent of web service technology has a significant
impact on the image processing. Because the image processing
tasks become more complex than before, image processing
software is increasingly large. The software also imposes harsh
restrictions on hardware conditions. Therefore, traditional
image processing methods have difficulty applying to resourceconstrained
mobile devices.
The above problem has been extensively explored;
however, the solutions only have been based upon centralized
computing model. For example, in order to adapt to the
resource-constrained mobile devices, the emphases are only
placed on the optimizing image processing algorithm and
improving the idea of software and hardware co-design
methodology. However, the corporation of hardware and
software make software loss of portability. Although a lot of
efforts are being spent on the problem of resource limitation in
mobile devices, the efficient method has yet to be developed.
Web service technology is characterized by a loosely
coupled, component-oriented [1], updated to maintain the
advantage of convenience, So many research studies and
applications have been carried out on this topic.
The aim of this paper is to put forward a web service-based
image processing method to solve the problem of inefficiency
of image processing in mobile devices. Web service technology
is a new distributed computing model. So it can make ful use of
heterogeneous network computing resources [2] and realize
resources sharing effectively. Exactly speaking for the method,
firstly image processing services’ registration, publication and
discovery can be completed in the server-sides. Secondly the
mobile clients only invocate specific services. So the mobile
clients can reduce the computing resources consumption.
However, the image processing effect was not affected any
longer.
In order to illuminate our program, this paper describes a
prototype system of image processing for the analysis and
discussion of the web service-based solution. Furthermore, a
kind of web service discovery algorithm is proposed to
improve the quality of image processing services.
II. RELATED WORK
Web service adopts a service-oriented architecture (services
oriented architecture SOA [3]), as shown in Figure 1. Service
providers are the owners of the services, which provide
platforms for services to access, and services to publish in
service registry center. Service requesters are the enterprises,
organization and individuals that have needs of specific service
functions. Service requesters can discover services and access
the information services binding in the registry center. Service
registry center is a database that can store service description
information; in the registry center, service providers publish
services, services requesters can discover and access binding
information.
So a web service-based prototype system consists of a
registry center, a service provider and service requestors. Each
of them is necessary and indispensable for the prototype system.
Exactly speaking, service provider can publish their services’
address and description in the registry center. Registry center
can provide services’ description information for the
requestors. Requestors can invocate the specific service by the
information the registry center offers.
978-1-4244-4131-0/09/$25.00 ©2009 IEEE
Figure 1. The framework of SOA
Web service as a new distributed computing model has the
following advantages:
Advantage of excellent encapsulation; they are the objects
or components on the web. The users can only see a list of
features objects provide.
Advantage of loose coupling; when the web services
change, the callers will not feel them. As to the service
requesters, as long as the service interfaces do not change, any
changes of services will make no influence on the service
requesters;
Advantage of standard protocols; all web services’ public
agreements must be described by open and standard protocols
such as SOAP [4], HTTP. Compared with the general objects,
they are more standardized and easier to understand by the
machine.
Advantage of integration capabilities; web service
technology adopts simple and easy-to-understand protocols to
express its principal which is described by the WSDL [5-7]. So
the technology fully shields the differences between different
software platforms. Not only EJB but also CORBA can easily
implement interoperability by standard protocols [8-9].
Considering the several advantages of web service, the
technology can be applied to mobile devices in order to
improve image processing efficiency.
III. DESIGN AND IMPLEMENTATION OF WEB SERVICE-BASED
IMAGE PROCESSING SYSTEM
A. Web service-based image processing model
Before the introduction of the system, two definitions are
given about image processing service.
Definition 1. Image processing service requests: image
processing request can be described as following two-tuple:
R= <K, C>
• R denotes image processing requests.
• K denotes R’s known parameters set. In this paper, Set
K is byte array of image files.
• C denotes R’s parameters constraint set. It defines the
constraints of image file’s format. Set C= {jepg, png,
bmp, gif, jpg}; the system support these formats of
image files.
Definition 2. An image processing web service can be
described as following four-tuple:
WS = <R, Q, Res, Qos>
• WS denotes image processing services names;
• R denotes image processing requests.
• Q denotes the entrance address and description of
specific image processing services.
• Res denotes image processing services’ memory usage.
• Qos denotes WS’s quality of services. In this paper,
Qos is related to image processing services’ response
time.
B. Architecture of the system
Web service-based image processing system consists of
three logic layers, as shown in Figure 2.
1) User interface layer: it is the implementation platform
of image processing system based on web service. Users can
use the wireless transmission technologies (WIFI,
BLUETOOTH, etc.) to process images through the image
processing interfaces.
2) Services management layer: its functions are image
processing services’ publication, discovery, and management.
The layer is able to ensure the services’ Qos.
3) Services implementation layer: the layer finishes the
specific image processing tasks. Image processing services are
distributed to various locations. They are regarded as "black
boxes" and provide a variety of interfaces but to hide the details
of the image processing.
C. Construction of the system
A web service can be seen as an object on the Web, the
system’s development is different from traditional software
development. It involves the web service’s publication,
discovery and invocation. The specific steps are as follows:
1) Construction stage: we completed definition and
description of the image processing service interfaces. The
system provided some service interfaces, such as anti-color,
gentle services, black-and-white and so on.
2) Services deployment stage: service interfaces and their
description are published to service registry center. UDDI
defines the storage paths of services interfaces’, which are
described by the WSDL. So the service requesters can use the
address to discover the image processing services interfaces.
3) Services discovery and invocation stage: the tasks of
services’ discovery and invocation are finished in the stage.
Firstly, the mobile clients set up the IP address of the service
registry center. Secondly, the mobile clients get service
address and service description from service registry center.
Finally, mobile clients invocate specific image processing
services.
Figure 2. The Framework of image processing system based on Web service
Figure 3. Data flow diagram of image processing system based on Web service
4) Services management stage: services security problems
are solved in the stage. The solution prevents the image data
from being maliciously destructed by third party.
D. Implementation of the system
Obviously, there is a great difference between web servicebased
and traditional imaging processing methods. The
traditional methods are always implemented by the centralized
computing model. However, the web service-based method is
based on distributed computing model. The specific steps are
as follows. The process of data flow diagram is shown in
Figure 3.
1) The mobile clients have to check the validity of
image files formats.
2) Picture files are compressed. Considering that transfer
rate of mobile devices is limited, the prototype system
has to compress image files in order to reduce the
transmission time.
3) Compressed picture files are converted into the flow
of bytes. And then, the mobile clients send the image
processing service inquiry requests to the service
registry center.
4) According to the customers’ requests, service
registry center searches the specific services. When
the appropriate services are discovered, services’
entrance address and description will be sent to
mobile clients.
5) Mobile clients invocate specific image processing
services by the services’ address and description.
When image processing tasks are finished, the service
providers will send results to mobile clients.
6) The mobile clients convert image byte stream into
image files.
E. Key algorithm-WSDA in the system
The system’s key is image processing services’ discovery.
According to the mobile clients’ requests, the image
processing services and their description should be discovered
by requesters immediately. Algorithms-WSDA is given as
Figure 4:
Let set R= {r1, r2...rm} be image processing service
requests. ResultSet is a set that denotes services discovered in
the service registry center. Let set P= {p1, p2…pn} be all kinds
of services in the service registry center. Let set pi= {s1,
s2…sk} (i=1…n) be a set of the services that finish the same
tasks. But there are some differences in details such as
parameters, Qos.
Figure 4. Algorithm-WSDA
IV. EXPERIMENT AND DISCUSSION
In order to test the viability and effectiveness of the
approach the paper proposed, we conducts a simulation
experiment. A laptop computer two small servers are used.
Two servers’ configurations are the same. Their configurations
are shown in table1.
TABLE I. CONFIGURATIONS TABLE
CPU Memory OS network
cards
Laptop 700MHz 128MB Linux 54 Mbps
Servers 2.6GHz 2GB Win2003 108Mbps
The laptop computer acts as a service requester. Two
servers act as service registry center and service providers
respectively. The mobile environment simulator Microsoft
Pocket_PC is utilized to test performance of two approaches.
One of them is based on web service (WSIPM) and the other is
based on centralized computing (CCIPM). According to basic
image processing functions (such as Sharpen, anti-color,
embossed, etc.), the above-mentioned approaches is compared
in the aspects of response time and memory usage. The results
are shown in Figure 5, 6.
0
500
1000
1500
2000
2500
gentle b-and-w sharpen anticolor
emboss
unit millisecond
WSIPM response time CCIPM response time
Figure 5. Comparison of response time
0
500
1000
1500
2000
2500
gentle b-and-w sharpen anti-color emboss
unit KB
WSIPM memory usage CCIPM memory usage
Figure 6. Comparison of memory usage
The experiment results shows that the web service-based
method can save an average of nearly 30% of the memory
usage, but response time was reduced by an average of about
25%. Two factors have been studied.
1) Web service-based image processing method
allocates complex image processing tasks to the
server with relatively high hardware configuration.
2) WSDA is polynomial time complexity. So the
specific image processing services could be
discovered rapidly in the service registry.
Thus mobile clients saved memory resources and improved
the processing efficiency. Especially for sharpening service, it
related to matrix multiplication and computational cost is
excessive. But web service-based approach showed more
obvious superiority.