Performance-based techniques for storing and retrieving images.

number: 
239
English
department: 
Degree: 
Imprint: 
Computer Science
Author: 
Haithem Kareem Al-Rubaie
Supervisor: 
Dr. Mohammed Ali Shalall
Dr.Riyadh A.K.Mahdi
year: 
1997
Abstract:

There are a number of techniques for representing pictorial information, array and raster representations are most frequently used lechmques. The amount of storage required to store images using such techniques is often excessive. So, in this thesis a number of image representation schemes are studied and developed including, hierarellical data .structures (quadtree and bintree) and Translation Invariant Data structure (TID), All these schemes arc compact spatial data structures that attempt to save storage by aggregating data having identical or similar values into blocks. For hierarchical scheme, different types of pointerless tree representations are implemented including: collection of leaf nodes and tree traversals representations. Also, a modified technique is proposed called qirudfrec largest nodes which is an improvement: over linear quadtree. All these poinlerless techniques require spaces less than pointer-based representations. A software package called Image Representation System ( IRS) was designed to store and retrieve 2-dimensional color and gray-level images using quadtree, bintree. and TID) techniques. In IRS the performance of these techniques are investigated and evaluated in terms of space saving and retrieval time. The minimum requirements to run IRS are IBM-AT or compatible computer, a VGA monitor, and mouse.