@ |
|||
Dr. Ramazan S. Aygün |
Research Areas |
||
|
|
||
|
Multimedia Databases
Although computer vision
and image processing techniques improved significantly, the subjective user
evaluation of video and image data as well as unsatisfactory object
recognition methods made it necessary incorporation of fuzzy, uncertain, and
imprecise data in multimedia databases. The multimedia database systems have
to enable information for an attribute at different levels of uncertainty and
support querying with fuzzy and crisp values independent of how data is
stored in the database. We have developed a conceptual database model for
multimedia database applications using ExIFO2 data model [15] and
used Fuzzy Object Oriented Model (FOOD) as the logical model. This multimedia
database is applied for videos of soccer games [3]. Multimedia Networking
The data whether stored in
the database or not has to be transmitted over networks shared by many users.
The congestion of the network and the compressed video data affect the
quality of presentations drastically. Especially, the loss of a frame does
not only affect the display of the frame, it also affects the display of the
dependent frames. We have proposed a Probe-loss utilization streaming (PLUS)
mechanism to reduce the loss of high priority frames [2]. For an MPEG video
stream, I frames have the highest priority and B frames have the lowest
priority. The status of the network is probed with B frames and I frames are
protected in a better way by reducing the congestion caused by the system. We
have built a client server middleware design strategy for streaming
multimedia presentations [1]. Interactive Multimedia
Presentation Management
The user interactions are
core elements of multimedia presentations. The loss and delay of the data
over networks complicates the synchronization if user interactions that
change the course of the presentations are enabled. The event-based or
constraint-based synchronization models provide powerful mechanism to handle
synchronization in distributed systems. However, the interactions like skip
and backward usually complicates the synchronization and the specification of
the synchronization. We have proposed a rule-based synchronization model
based on Event-Condition-Action (ECA) rules [10, 14]. These rules can be
specified by a rule definition language or can be extracted from SMIL
expressions [12]. The necessary constraints or rules for user interactions
are automatically extracted by the system and may be updated by the author. Model Checking
One major question is
whether the specified multimedia presentation is correct or not. A
synchronization specification may yield different results on different
presentation tools. It is important whether the presentation is actually
played correctly by the synchronization model. The specification properties
have also to be checked based on the synchronization model. The multimedia systems
are usually tested by simulation or by conducting experiments. Nevertheless,
in these cases it is not possible to cover all cases and sometimes mistakes
can be costly. We eventually applied model checking techniques on multimedia
synchronization. The synchronization rules are converted to PROMELA language
and the properties are verified by the SPIN model checker [4,11]. Consequently, the author is guaranteed to develop
correct presentations. Video Processing
Most
of the multimedia research is devoted on video. The video data challenges
researchers in compression, information retrieval, and transmission and led
many research areas. We have developed a novel video application to spatially
browse recorded video data [5]. The users are allowed to perform pan left and
right, tilt up and down, and zoom in and out, if the necessary information
can be extracted from various portions of the video. This requires high
resolution sprite (mosaic) generation, reduction of blurring in the sprite,
fast motion estimation and compressed domain processing to decrease
processing cost [6, 7, 8, 9]. We have developed a tool named as VideoCruise to spatially browse videos [5]. We have
tested this tool for distance education application. We have integrated video
browsing tool, synchronization model for user interactions and PLUS under a
multimedia system. Bioinformatics
Heterotrimeric G proteins interact with G protein-coupled receptors in response to
stimulation by hormones, neurotransmitters, chemokines,
and sensory signals to intracellular signaling cascades. We provide a robust
computational method for finding new G proteins and their homologs
using SVM
based pattern recognition algorithm. |
||||||||||
The views, opinions, and
conclusions expressed in this page are those of the author or organization
and not necessarily those of The
|