Content-Based Retrieval of Image

ü  System retrieves images that best match the description.

ü  Image content can be defined at different levels of abstraction

Lowest level

o   Image is a collection of pixels.

o   Rarely used.

The raw data can be processed to produce numeric descriptors capturing specific visual characteristics called features.

–     Color,

–     Texture, and

–     Shape

Features can be extracted from entire images describing global visual characteristics

                  Portions of images describing local characteristics

Highest Level

–     Images are often accompanied by metadata.

–     Metadata may contain information that cannot be obtained directly from the image, as well as an actual description of the image content.

–     Medical images.

Radiological report

Multidimensional Indexing

ü  Multidimensional descriptors

–     colour or

–     texture

ü  Search task can be made significantly more efficient by relying on multidimensional indexing structures.

ü  RDBMS should not support Multidimensional indexes.

Image Element representation = {Element name, Element colour, Element shape, Element attribute}.

By using this the images are retrieved

Image Indexing

The basis of image indexing is a semantic image representation. The total image representation is defined as: Global image representation and Image Elements representation,

where

Global image representation = {Image Name, Image Colour, Image Attribute}

And Image Elements representation includes the Image element representations and their places.

Image Element representation = {Element name, Element colour, Element shape, Element attribute}.

An Element is a small patch in the image with descriptive power. The best choice for an ensemble element is to form a set from which the image is composed.

The different elements in the semantic image representations are defined as:

1.      Name description.

Name of an image or elements in it could be given during the image entering.

2.      Attribute description.

It is borrowed from outside sources concerning the image elements. In the case of geographic maps, if the element represents a legend sign for a town, the user could enter the town population. The attribute description could be text or number.

3.      Colour description.

Each image voxel is considered as a point in a 3D colour space (Red, Green, Blue).