2. Data Modeling
• A data model is a formalism that helps specify the
aspects of the data relevant for their organization.
• Data modeling is the act of exploring data-oriented
structures. Like other modeling artifacts data models
can be used for a variety of purposes, from high-level
conceptual models to physical data models.
• Data modeling is conceptually similar to class modeling.
With data modeling you identify entity types whereas
with class modeling you identify classes. Data attributes
are assigned to entity types just as you would assign
attributes and operations to classes.
4. Traditional Data Modeling
• Traditional data modeling is different from class
modeling because it focuses solely on data.
• Class models allow you to explore both the
behavior and data aspects of your domain, with a
data model you can only explore data issues.
• Data modelers have a tendency to be much better
at getting the data “right" than object modelers.
6. Key Problem of Color Model
A color model is a quantitative representation of the colors that
are relevant in an application domain. For the applications that
involve human vision, the color model needs to represent the
colors that the human eye can perceive. Perceiving colors is
important for effective visual communication.
7. Bayesian Color Estimation for Adaptive
Vision-based Robot Localization
• The RoboCup domain poses some challenging problems for
state estimation, including moving objects, limited
computational power, and low resolution camera
information.
• To make use of colors, a robot has to map the raw color
values observed with its camera to the colors in the map.
• The key problem in this context is that the appearance of
these colors can change drastically under different lighting
conditions.
8. Key Color Priority Based Image
Recoloring for Dichromate
• People with color vision deficiency (CVD) have difficulty
to discriminate certain categories of color combinations.
• To deliver rich and distinguishable visual information for
the color blind people, a natural way is to re-color the
images.
• The key problem in this context is how to efficiently
re-organize the color information thus becomes a critical
issue.
9. Color Base Retrieval and Recognition
• As the world enters the digital age, visual media is
becoming prevalent and easily accessible. Factors
such as the explosive growth of the World Wide
Web, terabyte disk servers, and the digital versatile
disk, reveal the growing amount of visual media
which is available to society.
• Two key problems in color indexing are
determination of color space and finding the best
distance measure.
• Most of the attention has been focused on the color
model with little or no consideration of the noise
models.
10. A Skin Color Model Based on Modified
GLHS Space for Face Detection
• Face detection is a very important key step in fully
automated face recognition systems. The purpose of
face detection is to determine the locations and sizes of
human faces in images.
• Face detection has been successfully used in
biometrics, video surveillance, human-computer
interface and image database management.
• However, since the color of facial pixels is sensitive to
different illumination conditions, it is hard to achieve
stable detection performance. Therefore, skin color
based detection methods have limitations in practice.
11. Shortcomings of Traditional Data Modeling
Data Storage
• Data Security
The data stored in the flat file(s) can be easily accessible and
hence it is not secure.
• Data Redundancy
In this storage model, the same information may get duplicated
in two or more files. This may lead to higher storage and
access cost. it also may lead to data inconsistency.
• Data Isolation
Data Isolation means that all the related data is not available in
one file. Usually the data is scattered in various files having
different formats. Hence writing new application programs to
retrieve the appropriate data is difficult.
12. Shortcomings of Traditional Data Modeling
Data Storage
• Program/Data Dependence
In traditional file approach, application programs are closely
dependent on the files in which data is stored. If we make any
changes in the physical format of the file(s), like addition of a data
field, all application programs needs to be changed accordingly.
Consequently, for each of the application programs that a programmer
writes or maintains, the programmer must be concerned with data
management. There is no centralized execution of the data management
functions. Data management is scattered among all the application
programs.
13. Shortcomings of Traditional Data Modeling
Data Storage
• Lack of Flexibility
The traditional systems are able to retrieve information for
predetermined requests for data. If we need unanticipated
data,
huge programming effort is needed to make the information
available, provided the information is there in the files. By the
time
the information is made available, it may no longer be required
or
useful.
14. Shortcomings of Traditional Data Modeling
Data Storage
• Concurrent Access Anomalies
Many traditional systems allow multiple users to access and update
the same piece of data simultaneously. However this concurrent
updates may result in inconsistent data. To guard against this
possibility, the system must maintain some form of supervision.
But supervision is difficult because data may be accessed by many
different application programs and these application programs may
not have been coordinated previously.
15. Shortcomings Of Modeling Media Data
With Traditional Data Modeling
Relational Model
• The constraints underlying the database in terms of a set of first-order
predicates, defined over a finite set of predicate variables.
• Entity relationship model are limited in what they can express. It helps to be
aware of the limitations up front, since it can affect on represent the
enterprise of the modeling that chosen.
A simple relational database with two relations
16. Shortcomings Of Modeling Media Data
With Traditional Data Modeling
• Performance: A major constraint and therefore
disadvantage in the use of relational database system is
machine performance. If the number of tables between
which relationships to be established are large and the
tables themselves effect the performance in responding
to the sql queries.
• Slow extraction of meaning from data: if the data is
naturally organized in a hierarchical manner and stored
as such, the hierarchical approach may give quick
meaning for that data.
17. Shortcomings Of Modeling Media Data
With Traditional Data Modeling
• Physical Storage Consumption: With an
interactive system, for example an operation like
join would depend upon the physical storage also.
It is, therefore common in relational databases to
tune the databases and in such a case the physical
data layout would be chosen so as to give good
performance in the most frequently run
operations. It therefore would naturally result in
the fact that the lays frequently run operations
would tend to become even more shared.
18. Shortcomings Of Modeling Media Data
With Traditional Data Modeling
Object Oriented
• The disadvantages of object oriented model or
programming is that is larger than other types of
programming. It also requires a lot of work even before
the first piece of code is written, so labor is costly. They
require an abundance of system resources to run the
program and due to their size they are slower than other
programs.
• Larger program size: Object-oriented programs typically
involve more lines of code than procedural programs.
19. Shortcomings Of Modeling Media Data
With Traditional Data Modeling
• Steep learning curve: The thought process involved in
object-oriented programming may not be natural for
some people, and it can take time to get used to it. It
is complex to create programs based on interaction of
objects. Some of the key programming
techniques, such as inheritance and
polymorphism, can be challenging to comprehend
initially.
• Slower programs: Object-oriented are typically slower
than procedure-based programs, as they typically
require more instructions to be executed.
20. Shortcomings Of Modeling Media Data
With Traditional Data Modeling
• Not suitable for all types of problems: There
are problems that lend themselves well to
functional-programming style, logic-
programming style, or procedure-based
programming style, and applying object-
oriented programming in those situations will
not result in efficient programs.
21. Shortcomings Of Modeling Media Data
With Traditional Data Modeling
Object-Relational Models
• This approach has the obvious disadvantages
of complexity and associated increased costs.
Further, there are the proponents of the
relational approach that believe the essential
simplicity and purity of the relational model
are lost with these types of extension.
22. Shortcomings Of Modeling Media Data
With Traditional Data Modeling
• Object-Relational models vendors are
attempting to portray object models as
extensions to the relational model with some
additional complexities. This potentially
misses the point of object
orientation, highlighting the large semantic
gap between these two technologies. Object
applications are simply not as data-centric as
relational-based ones.
34. The origin of the RGB model corresponds to the
lack of colour signal; ex:black while the
diagonal of the origin corresponds to maximum
signal levelsThis model is commonly implemented using
data structures that allocate
same number of bits to each colour channel
35. For example, a 3-byte representation of
colour can represent 24 bit different
colour instances,
would allocate 1 byte each to each
colour channel thus distinguish 256
intensities of pure red, green and blue
called pallete
36. An image is represented as 2-D
matrix, each cell contains 24-bit
colour instance. The cell is called
as pixelEx:
representation
of an image of
1000x1000
requires
24x1000x1000
bytes. When the
space required is
not But…how?
37. First method is by reducing the precision
of the colour channel by allocating 4
bits per colour channel instead of 8 bits
Hence there will be 4096 different
colour instances instead of 16,777,216
colour instances
38. Second method is by using the colour
table. Colour table is a lookup table
that maps from a less precise colour
index to a more precise colour instanceAssume that we
can process all
the pixels in an
image to identify
the best 4096
distinct 24-bit
colours.
These colours are
put into the
lookup table. For
Hence, when we
want to display a
photo, the
software use the
lookup table and
convert the
colour indexes to
actual 24-bit RGB
Example of
lookup table
39. Third method is by using the YRB, YUV and
YIQ model. Human eyes are more sensitive
to contrast than to colours.
Hence, colour model that represents
grayscale as explicit component is more
effective than combination of RGB in
creating reduced representationsYRB
Luminance = amount
of light (y) in RGB-
based colour
instance is :
y=0.299R+0.587G+0.11
4B
Shows that the blue
colour is the least
contribute to the
perception of light.
Given the luminance(y)
and the existing RGB
channels; R and B, a
new colour space YRB
40. YUV
Subtract the luminance component from
the colour components:
U = 0.492(B-Y)
V = 0.877(R-Y)
These equations ensure black and white
pictures have no R and B components. U
and V components reflect the
chrominance of the corresponding
colour instance
YIQ
For further reduction of bit
representation, rotate U and V