Multimedia Database

Design and Development

MMS 144 – Principles of Multimedia Information Management

Revision History
April 2024: Main content= – Ria Borromeo
Outline
1. Introduction
2. Challenges of Multimedia Database Design
3. Approaches to Multimedia Database Design
4. Multimedia Database Design Stages
5. Multimedia Database Development
6. Best Practices
7. Summary
8. Case Study

2
Introduction
Why do we have to study database design specifically for
multimedia?

In today's world, multimedia data – including images, audio, video, and

text – is a crucial component of information storage and retrieval.
Traditional databases, designed for structured data like text, struggle to
efficiently manage these complex multimedia objects.

3
Challenges of Multimedia Database Design
• Large Data Size: Multimedia data can be significantly larger than
traditional text data, posing storage and retrieval challenges.
Images, audio, video take up large storage space!

• Data Complexity: Multimedia data is often unstructured or semi-

structured, making it difficult to represent and search.

4
Challenges of Multimedia Database Design
• Content-Based Retrieval: Unlike text databases where keywords
suffice, multimedia retrieval often requires searching based on
content, like color in an image or melody in audio.

• Performance: Efficient retrieval of multimedia data requires

optimized indexing and search algorithms.

5
Approaches to Multimedia Design
Linked Multimedia Databases
Object-Oriented Multimedia Databases (OOMDBs)
Hybrid Database

6
Linked Multimedia Databases
• Store the actual multimedia data (images, videos) separately on a
file system.
• Create a database table containing metadata (descriptive
information) about the multimedia data.
• Link the metadata table to the file system location using
identifiers.

7
Linked Multimedia Databases
Advantages
• Efficiency – storing only metadata in the database reduces storage
requirements, especially for large files like videos
• Flexibility – the database schema can be easily modified to
accommodate new metadata fields without affecting the stored
multimedia data
• Performance – querying and retrieving metadata is typically faster
vs. accessing large multimedia files within the database itself
• Scalability – the database can easily scale to accommodate a
growing collection of multimedia data since the files are stored
separately

8
Linked Multimedia Databases
Implementation Concerns
• File System Management
• Proper organization and management of the file system where the
multimedia data resides must be ensured.

• Data Integrity
• Robust mechanisms are needed to ensure that the links between the
metadata and the actual multimedia data are valid.

9
Linked Multimedia Databases
When to use Linked Multimedia Databases?
• Large multimedia collections
• Performance is critical
• Scalability is required

10
Linked Multimedia Databases
Examples:
• Digital libraries: managing collections of images, audio recordings,
and even video documentaries;
• Multimedia e-commerce platforms: storing product images,
videos, and descriptions for online sales;
• Social media platforms: handling user-uploaded multimedia
content like profile pictures and video posts

11
Object-Oriented Multimedia Databases
• Model multimedia data as objects with attributes and methods.
• Objects closely resemble real-world entities so it is intuitive to model
complex multimedia content.
• An image object can encapsulate image data itself along with its
descriptive metadata and manipulation functions.
• Allow encapsulating data and functionality within a single unit.
• By bundling data and methods within objects, OOMDBs promote data
integrity and security. Only authorized methods can access and modify
the multimedia data.

12
Object-Oriented Multimedia Databases
When to use OODBMS?
• When you need to manage intricate multimedia objects with
diverse data types and functionalities
• When the application requires searching and retrieving
multimedia data based on its content (e.g., searching for images
based on color or video based on audio characteristics).
• When you anticipate significant growth in your multimedia data
volume and complexity.

13
Object-Oriented Multimedia Databases
Examples:
• Digital Asset Management Systems: storing and managing various
multimedia assets like images, videos, and audio for creative
professionals.
• Medical Image Databases: organizing and retrieving medical
images like X-rays and MRIs for diagnosis and research purposes.
• Multimedia Learning Systems: creating interactive learning
experiences that incorporate multimedia elements like images,
videos, and simulations.

14
Hybrid Multimedia Databases
• Combine elements of Linked and Object-Oriented
approaches
• Provides versatile approach for managing complex
multimedia data

By combining the strengths of linked and object-oriented databases, they

provide a balance between efficiency, flexibility, and content-based retrieval
capabilities.
However, the increased complexity of design and implementation needs to be
carefully considered before choosing this approach.

15
Database Design Steps for
Multimedia
1. Data Modeling
2. Media Modeling
3. Schema Design
4. Media Access and Query
Processing

16
Database Design Steps
1. Data Modeling
• Identify multimedia data types (images, audio, video) and
their associated metadata (e.g., size, format, creation
date)
• Define relationships between multimedia objects and
other data (e.g., text descriptions, user annotations).
• Use ERDs to visualize relationships

17
Database Design Steps
2. Media Modeling
• Determine how to store the actual multimedia data
• Options include storing raw data, using compressed formats, or
employing feature extraction techniques to represent the content

3. Schema Design
• Design the database schema, which defines the structure of the
tables of their relationships
• Consider data types, constraints, and indexing strategies for
efficient retrieval

18
Database Design Steps
4. Media Access and Query Processing
• Develop mechanisms to access and manipulate multimedia data
Think about how you will access your data.
What are the queries you will need?

• Implement content-based retrieval techniques using features extracted from

the data
Things to consider:
Are you going to retrieve images based on color? Texture? Shape?
Are you going to retrieve videos based on a specific motion (parked
car, moving person) or based on color or scenes?

19
Multimedia DB Development
This phase involves creating the database schema (structure) within
the chosen DBMS and populating it with data.

20
Best Practices In Multimedia DB Design
• Standardization: use standard formats for multimedia data (e.g., PNG
for images, MP3 for audio) to ensure compatibility and future-
proofing.
• Data Compression: utilize compression techniques to reduce storage
requirements and improve retrieval speed.
• Indexing: implement efficient indexing mechanisms for both
metadata and extracted features to facilitate content-based searches.
• Scalability: design the database to handle future growth in data
volume and diverse multimedia types.

21
Summary
Designing databases for multimedia presents unique
challenges, but by following these approaches and best
practices, you can create a robust and efficient system
for managing and retrieving your multimedia data.

22
 Case Study
Designing an Earth Sounds Database System

23
Earth Sounds Database System
An Earth Sounds Database System is a specialized type
of multimedia database designed to collect, store,
manage, and retrieve audio recordings of natural
phenomena that produce distinct sounds.
Examples: sounds from volcanic eruptions, earthquakes,
winds passing through rock formations, sand dunes, or any
other natural occurrence where the environment creates
unique acoustic phenomena.

24
Earth Sounds DB System
Purpose
• To preserve the sounds of the Earth as a cultural and scientific
resource
• To enhance learning about geology, geography, and environmental
science
• To engage the public in environmental and geological issues by
making the sounds accessible

25
Earth Sounds DB System
Features
• Supports various media types including audio, images, and video
• Provides detailed metadata for each media file, such as format, size, creation
date, and location
• Allows users to add annotations to recordings
• Features sophisticated search functionalities that allow both metadata-based
and content-based retrievals
• Ensures robust data management and scalability

26
1. Data Modeling
Identify multimedia types and metadata
• Audio: Format (WAV, MP3), Length (seconds), File Size (MB),
Recording Date, Quality (bitrate)
• Images: Format (JPEG, PNG), Resolution (pixels), File Size (KB),
Creation Date
• Video: Format (MP4, AVI), Length (minutes), Resolution, File Size
(MB), Creation Date

27
1. Data Modeling
Identify entities
• Site: Represents locations where sounds are recorded.
• SoundEvent: Each natural sound occurrence, linked to a Site.
• Recording: Media files capturing the sound event. Could be audio,
images, or videos.
• User: Users of the system who may annotate or tag recordings.

28
1. Data Modeling
Identify relationships
• A Site can have multiple SoundEvents.
• A SoundEvent can generate multiple Recordings (polymorphic:
audio, images, video).
• Users can annotate many Recordings.

29
1. Data Modeling
Simplified (ERD) – attributes not included yet.
1 n
Site has SoundEvent
1

generates

n
n n
User annotates Recording

30
2. Media Modeling
Determine storage methods for multimedia data
• Audio and Videos: Store externally as files in a file system and
save the path in the database due to large size.
• Images: Depending on size, can be stored directly in the database
as BLOBs or externally like audio and video.

31
3. Schema Design
Database Tables
• Sites: SiteID, Name, GeographicCoordinates, Description
• SoundEvents: SoundEventID, Description, Date
• Recordings: RecordingID, MediaType (audio, image, video), FilePath,
Format, Length, FileSize, CreationDate
• Users: UserID, Firstname, Lastname, Username, Email
• Annotations: AnnotationID, RecordingID, UserID, Text, Timestamp
• Sites-SE: SiteID, SoundEventID
• SE-Recordings: SoundEventID, RecordingID
• User-Recordings: UserID, RecordingID

32
4. Media Access and Query Processing
Queries Needed
• Retrieve all recordings for a specific sound event.
• Find all annotations for a recording
• Search recordings by format, length, or file size

Content-based retrieval technique

• Extract and index features such as frequency or duration from
audio recordings for advanced search capabilities

33
DBMS Development
Create tables
• Write SQL code to create the tables
Populate tables
• Insert sample data in the tables
Create sample queries

Check all queries here:

https://onecompiler.com/sqlite/42c5ej9bh

34
DBMS Development
Create tables
• Write SQL code to create the tables
Populate tables
• Insert sample data in the tables
Create sample queries

Check all queries here:

https://onecompiler.com/sqlite/42c5ej9bh

35
Summary
We designed an Earth Sounds Database and implemented it using
SQLite. It could store various sounds observed in the earth, which users
can annotate. We aim to store more data, test more quries, and share
the database publicly in the future.

36