A DATABASE MANAGEMENT SYSTEM KERNEL

FOR IMAGE COLLECTIONS

Liana Stanescu, Dumitru Burdescu, Cosmin Stoica and Marius Brezovan

University of Craiova, Faculty of Automation, Computers and Electronics

Keywords: Multimedia database management system, content-based visual retrieval, color histogram, medical imagery.

Abstract: The paper presents a single-user, relational DBMS kernel, for managing visual information. The functions

of this multimedia DBMS are: creating/deleting databases and tables, adding constrains, inserting, updating,

deleting records, text based querying and content based visual querying using the color characteristics. The

originality character of this DBMS is given by two aspects: the first aspect refers to the Image data type that

permits binary storage of images and extracted color information represented by the color histogram with

maximum 166 colors; the second aspect refers to the visual interface for building content-based image query

using color characteristics, that generates a modified SELECT command that will be sent to kernel for

execution. This DBMS has as advantages the low cost and easiness in usage, being recommended in

medical or art domains where large amounts of visual information are collected, managed and queried.

1 INTRODUCTION

Multimedia enhances in quality and quantity the sent

information. That is the reason why multimedia and

imaging represent an important component in

computing. Multimedia information is complex,

needs a lot of disk space and operations as updating,

concurrent access and elements searching. The best

solution to store, manage and find multimedia

information is to use a multimedia database

management system. A MMDBMS must have a

series of important characteristics: support for

multimedia data types, possibility to manage large

number of multimedia objects, hierarchical storage

management, conventional database capabilities and

information retrieval capabilities (Khoshafian and

Baker, 1996).

In conclusion, MMDBMS must have complex,

high level interfaces for browsing and querying the

multimedia objects.

Browsing and navigation is a technique where

the user finds the object he wants to see or to update,

starting from higher-level root objects. Such a

technique is used in CAD/CAM/CASE applications,

or in hypermedia documents databases (Khoshafian

and Baker, 1996). This technique is not always

adequate because the user might want to find

associated objects, namely objects that have in

common certain characteristics or attributes. The

user specifies the query using a query language or a

visual query tool, and then it is sent to the

MMDBMS for processing. The query may relay on

multimedia objects attributes or can be content-

based (Khoshafian and Baker, 1996).

The paper presents a single-user, relational

DBMS kernel that supports managing medium sized

collections of digital images. For this, besides the

usual data types (int, char), the proposed DBMS

allows storage of images using Image data type.

Excepting traditional operations in a database, the

DBMS offers a visual interface for building, in an

interactive manner, the content based visual query,

using the color characteristics. To do this, when the

user inserts an image in the database, besides storing

the image in binary format, the algorithms for

automatic extraction of the color information are

executed. The RGB color space is transformed in

HSV color space and quantified to 166 colors. The

histogram represented by 166 colors, is also stored

with the binary image in the database.

The proposed kernel for DBMS is original in the

way it manipulates the images and because it gives

the possibility for content based retrieval using color

characteristics. It is easy to use in areas that manage

not only traditional information (numbers and

strings) but also visual information (medicine, art).

The benefits of the content-based visual query in

several areas of the medical domain were

530

Stanescu L., Burdescu D., Stoica C. and Brezovan M. (2007).

A DATABASE MANAGEMENT SYSTEM KERNEL FOR IMAGE COLLECTIONS.

In Proceedings of the Ninth International Conference on Enterprise Information Systems - DISI, pages 530-533

DOI: 10.5220/0002406805300533

 SciTePress

emphasized (Muller et al, 2005). The presented

DBMS has a low cost and might be an alternative to

high level but also very expensive solutions like:

Oracle 10g and Intermedia that support all types of

multimedia; or MS SQL Server with Image data

type, but in this case a complex application for

managing visual information is needed (Chigrik,

2007, Oracle, 2005).

2 THE FUNCTIONS OF THE

DBMS KERNEL

Next, the functions of the software tool will be

presented.

2.1 Database Management

To create a new database, the user must specify the

name of the database in the dialog window. First, the

created databases are empty. For each database there

will be made a new folder in the ”Databases” folder,

and all the pieces of information will be stored in the

new folder. The name of this folder is the name of

the database. After creating a database, it will be

listed in the tree on the left side of the application

window. This tree is used to see all databases,

including their tables.

To eliminate a database, it must be selected from

the tree on the left side of the window. Before

completing the delete, the user must confirm the

action because the database will be deleted entirely,

including the database folder and all the files.

2.2 Table Management

To create a table, the user must select a database.

Also, he must specify the name of the table, the

columns, primary and external keys if any. The

names of the tables in a database are unique. For

each table it will be created a new file with a specific

structure (a header area and a recording area). This

file is created in the database folder having the name

of the table and the “.tbl” extension. The user has to

specify the structure of the table: the columns, data

types, size, applicable constrains. The name of a

column has to be also unique. This aspect is ensured

by the DBMS.

Three types of data are implemented: int, char

and image. For the fixed length strings, the user

specifies the maximum size. A new type of data is

introduced – Image. It permits storing in the

database an image having one of the following

formats: bmp, gif or jpg. When creating a new table,

the user may also specify the primary key. It can

include one or several columns.

The user may specify a 1:m connection between

two tables: a parent table (on the 1 side of the

connection) and a son (on the m side of the

connection). For this it must be used the Foreign

Keys tag, in the same window. The user easily

chose the parent and son table and the foreign key.

The primary key and the foreign key must have the

same type and the same size. If there is a connection

between two primary keys, the connection will be

1:1. The structure of the table might be seen at any

moment in the main window of the DBMS, using

Components tag. Once the table is created, the user

can add new records, modify or delete existing ones,

using the record editor.

2.3 Updating Data in Tables

The user can add a new record only if the previous

record was correctly added and saved in the

corresponding file of the table. A record is correct if

all the fields are filled with information having the

type described in the structure. If one of the table’s

fields has the Image data type, when inserting, the

“Chose image” dialog window is opened to permit

the user to choose the image he wants to add.

In content-based visual query on color feature

(the color is the visual feature immediately

perceived on an image) the used color space and the

level of quantification - meaning the maximum

number of colors - are important. The color

histograms represent the traditional method of

describing the color properties of the images. They

have the advantages of easy computation and up to

certain point are insensitive to camera rotating,

zooming, and changes in image resolution (Del

Bimbo, 2001, Smith, 1997). In this DBMS the

images are represented in HSV color space because

of its properties: uniformity, completeness,

compactness and naturalness (Smith, 1997),

properties which make the space good for utilization

in the process of the content-based visual retrieval.

The transformation from the RGB color space to

HSV color space is nonlinear, but still easy to

implement (Smith, 1997).

Quantifying operation is needed to reduce the

number of colors used in content-based visual query

from millions to tens. The solution proposed by J.R.

Smith was chosen, the one referring to quantification

of the HSV space that produces a compact set of 166

colors (Smith, 1997).

The effectuated studies on nature and medical

images have shown that the use of the quantified

HSV color space to 166 colors is one of the best

choices in order to have a content-based visual query

A DATABASE MANAGEMENT SYSTEM KERNEL FOR IMAGE COLLECTIONS

531

process of good quality (Smith, 1997, Stanescu et al,

2006). So, when the user inserts an image in the

database, all these operations (color space

transformation and quantization) are executed. In the

column defined as Image type, the binary image and

the color histogram are stored.

Regarding the delete operation, a record can be

deleted only if it was selected before in the table

editor. The user selects the line corresponding to the

record he wishes to delete then he uses the delete-

record option.

Each column in a record can be updated using

the table editor. For updating, the user has to select

it, namely to put the mouse pointer in the

corresponding cell. When he moves the pointer in

another location, all the updates will be saved in the

table file.

2.4 Content-Based Image Query on

Color Feature

Content based retrieval implies selection of an image

as a query image and finding all the other images in

a database that are more alike query image. This

process can be executed at image level (CBIQ) or

using only certain query regions of the image and

considering their relative or absolute localization

(CBRQ). Content-based visual query is mainly

based on extracting image characteristics: color,

texture, shape (Del Bimbo, 2001, Faloutsos, 2005).

The DBMS presented here gives the possibility to

create content based visual queries based on color, at

image level, in a simple manner, using the Query

window that is presented in figure 1. The elements

of this window are:

− From – the table used by the query is chosen from

the list with the names of the tables

− Select – allows the user to select the field or fields

that will be listed in the result query

− Similar With – opens the dialog window for

selecting query image

− Where – specifies the Image type column on

which content-based image query is applied

− Comparing method – there are few types of

methods to calculate the similarity between the

query image and the target images in the

database: histogram intersection, Euclidian

distance and quadratic distance between

histograms.

The effectuated studies on medical images have

shown that each of the methods listed above

provided closely results in the content-based visual

query. The studies have also shown that the results

are complementary to each other (Stanescu et al,

2006). For example, if the doctor makes the same

query several times, choosing different methods, he

will obtain more relevant images.

− Threshold – the user can specify a threshold for

image similarity. Under this threshold, the

resulted images are listed

− Maximum images – maximum number of images

returned by the query process

Figure 1: The window that permits the building of the

content-based image query.

After building this query, a modified SQL

SELECT query is obtained, adapted to content-based

image query. The structure of the command is:

Select patients.diagnosis,

patients.img From Patients where

Patients.img Similar with Query

Image (method: Histogram

Intersection, max.images 5)

This modified Select command specifies that the

result set will be obtained from Patient table,

considering the values from diagnosis field and

similar images with the query image. The method

used for calculation is histogram intersection and

there will be listed only 5 images. The results will

include also the distance of the similarity between

the query image and the result image.

2.5 Data Organization

In the application folder there will be automatically

created the folder named “Databases”. When

creating a database, a new folder is created with the

name of newly created database. All the files for the

database will be stored here. Every table of the

database is stored in a separate file having the ‘.tbl’

extension. This file has two components: a header

and a data area. The header is added when the user

creates the structure of the table, and the data area,

when the user inserts, updates or deletes records.

ICEIS 2007 - International Conference on Enterprise Information Systems

532

The header will contain the following information

regarding the structure of the table:

- Number of records for header

The header will contain a record for each column

in the table. It will also contain a record for primary

key of the table and a record for each external key.

- Size for each record in header (a header record

contains data about a column in table: name, type

and length in case of character string; or about

primary key; or about foreign key/keys).

- Header records

For an image, in the data area of the file, the

DBMS stores the following information:

- image type (bmp, jpg or gif);

- height and width of the image;

- number of bytes needed to store the image;

- the image in binary format;

- 166 integer values representing color histogram.

A series of methods frequently used in the

medical domain are also implemented: rotating,

zooming, pseudo-colors, the similarity distance

between two images, a thumbnail representation, etc.

The Image data type is generally in compliance with

the SQL/MM standard.

3 CONCLUSIONS AND FUTURE

WORK

The paper presents the functions and the data

organization (file header and record area) of a

DBMS that has as main goal the improving of the

visual information management. In order to realize

this goal, the DBMS has a graphical interface for

designing the content-based visual query on color

feature. The image color information is represented

by color histogram with 166 values in HSV color

space. The histograms intersection, Euclidian

distance and quadratic distance between histograms

can be used for computing the similarity between the

query and the target images. The presented DBMS

uses the Image data type that permits storing the

image in binary mode and other necessary data

(image type, dimensions, color histogram). It has a

low cost and can be easily used in any domain that

manages images. For implementation, Java

technology was used.

The MMDBMS was tested using a system with

the following characteristics: AMD Athlon 3000+

processor, 1 GB RAM Memory, 2x150Gb RAID

HDD, Windows XP Professional operating system.

The effectuated experiments showed that the time

necessary for content-based image query process or

for displaying the records (this process implies the

extracting and viewing the binary images in the

database) is good. For example, in the case of 1000

records, the query time is 0.89 s and the display time

is 15.28 s; in case of 20000 records, 4.7 s and 73 s

respectively.

To enhance the quality of the software tool, the

following directions will be searched:

- the disk space management, taking into account

that the multimedia data needs a lot of space

- studying and implementation of traditional

indexing algorithms and specific algorithms for

spatial indexing

- adding new types of traditional or multimedia

data, such as video or DICOM, taking into

consideration that the medical domain is the

main target for this MMDBMS

- studying and implementation of the concurrent

database access

- extending content based retrieval combining

color and texture characteristics

REFERENCES

Chigrik, A., 2007. SQL Server 2000 vs Oracle 9i.

http://www.mssqlcity.com/Articles/Compare/sql_server_v

s_oracle.htm

Del Bimbo, A., 2001. Visual Information Retrieval,

Morgan Kaufmann Publishers. San Francisco USA.

Faloutsos, C., 2005. Searching Multimedia Databases by

Content. Springer.

Khoshafian, S., Baker, A.B. 1996. Multimedia and

Imaging Databases. Morgan Kaufmann Publishers,

Inc. San Francisco California

Muller, H., Rosset, A., Garcia, A., Vallee, J.P.,

Geissbuhler, A., 2005. Benefits of Content-based

Visual Data Access in Radiology. Radio Graphics.

25:849-858

Oracle, 2005. Oracle InterMedia: Managing Multimedia

Content.

http://www.oracle.com/technology/products/intermedi

a/pdf/10gr2_collateral/imedia_twp_10gr2.pdf

Smith, J.R., 1997. Integrated Spatial and Feature Image

Systems: Retrieval, Compression and Analysis, Ph.D.

thesis, Graduate School of Arts and Sciences.

Columbia University.

Stanescu, L., Burdescu, D.D., Ion, A., Brezovan, M.,

2006. Content-Based Image Query on Color Feature in

the Image Databases Obtained from DICOM Files. In :

International Multi-Conference on Computing in the

Global Information Technology. Bucharest. Romania

A DATABASE MANAGEMENT SYSTEM KERNEL FOR IMAGE COLLECTIONS

533