×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Proceedings of the Symposium on Progress in Data Processing and Analysis

71
A Fast Storage Unit With Integrated Database
Processor for Image Data
Lieckfeldt, P.; Missling, K.-D. ; Neumann, B.
Academy of Sciences of the GDR
Institute for Space Research
Satellite Ground Station
Kalkhorstweg
Neustrelitz
2080
Abstract
This paper gives a survey of an efficient image storage system with integrated data
base functions.
Storing the physical image data on external media causes mainly two problems: stor
age capacity and access time. An intelligent disk controller for disk arrays which tack
les both problems will be presented.
The management of images can be done in easy cases with a conventional DBMS. Ex
tending such a DBMS by a specialized search hardware to a DB Machine allows a fas
ter image retrieval. A higher level of image retrieval demands a semantic description
of the image content.
1. Introduction
Images are a very important source of information in many fields (e. g. remote sens
ing, meteorology). In nearly all cases images and related information must be stored
and managed. A concept for an Image Database Management System (IDBMS) has to
offer solutions for the following problems:
• On which media the huge amount of data (e.g. series of raster data, maps)
should be stored? How is it possible to guarantee a required data transfer rate?
• How is it possible to access to the image information fast and in a user-friendly
manner?
For applications with extreme time requirements we concepted an image storage unit
with high storage capacity and data transfer rate (in MByte/s range). The image data
management is done by an integrated IDBMS which allows a very quick access to
image annotations. In chapters 2 and 3 a detailed description of both components
(storage unit, IDBMS) is given.
2. Storing of physical images
If short I/O-service time is a goal implementing an IDB then direct access storage de
vices (DASD) as hard disks or optical disks are often used as storage medium. They
have a sufficient cost per bit ratio , an access time in the milisecond range and an ac
ceptable storage security. The data transfer rate is in the range of some MByte/sec.
If demands for retrieval are very high the so called "access gap" between primary
memory and disks limites the system performance. Because there isn’t any new stor