71 
To manage and retrieve satellite imagery data, image data is 
considered as a Binary Large Object (BLOB) multimedia type 
data. Multimedia type data has it own data management and 
query system. A Standard Database Management Sytem 
(DBMS) can perform exact matching query (using SQL) from a 
database tile. A Spatial Database can perform suitability 
analysis using spatial algebra. An image database has a special 
query type data retrieval called similarity (Guivada, 1995). 
3. REMOTE SENSED DATA MANAGEMENT 
Two major issues about variation in data management will be 
discussed. Metadata is a brief description of image content, 
format and contents that depends on each user. Image data 
format is a binary data format that allows the storing of multiple 
layer information. 
Content Based Information is a way of information handling on 
image file. QB1C (Query By Image Content) uses color, texture 
and shape to retrieve image similarity. One content based 
approach is Entity Relationship Diagram (ER-D) of Picture 
(C.Yu, 1994). Figure 1 show an example of ER-D of a picture. 
Figure 1 
Example ER-Diagram of Image / Picture Data (C.Yu, 1994) 
Retrieving Image data using Spatial and Temporal Data has 
been introduced by Mohammad Nabil (1997), replacing text 
based metadata with content based information. The model that 
introduced by Nabil is 2 Dimension Projection Interval 
Relationship (2D-PIR), which uses a directed graph data 
structure to represent network of objects and their spatial 
relationship. 
2D-P1R model use a projection of an image into a set of 
polygon vector that reflects shape and position of each 
important part (content) of image. The process to create a shape 
polygon from raster image can be a manual sketch or automatic 
shape edge detection using Artificial Intelegence technology. 
From this vector, 2D-PIR model generates a spatial relationship 
that record relationship among vectors (topology) in one image, 
started from upper-left side of image. The format of this record 
is a linked list that records list of objects which has spatial 
location on the right side of referenced object. Figure 2 show a 
simple sketch of 3 object from an image data (A, B and C) and 
its 2D-PIR linked list scheme. Each record in linked list contain 
polygon code and its spatial relationship code. 
3.1 Metadata and file format 
Current image data handling is organised by metadata or a 
header file in ASCII text format. Metadata files contain 
important information about what is stored in an image file, and 
how the data is stored (file format). Some example of important 
information that stored in metadata files are : Location 
Information, Coordinate System and boudaries, Projection 
Information, Sensor type, Acquisition date and description of 
image content. 
Metadata (in the above format) , helps user to make a quick 
assesment of the image data before actually retrieving the image 
data file. Its content is depend on user who create the metadata. 
The interpretation of metadata is also a manual interpretation 
process, and is followed by manual image data retrieval. 
3.2 Data independence 
Current image data handling varies depending on remote 
sensing software. Although Remote Sensing software can read 
many types of Image data format (BIP, BIL, BSQ, or original 
format of Satellite Imagery or), each software use their own 
proprietary format to store and manipulate image file. In ER- 
Mapper, simple Image data management can be implemented 
using Algorithm file. We can create an algorithm that handle 
multiple image surface, example of ER-Mapper algorithm is 
shown in figure 3.1. 
C 
A 
B 
Figure 3.1 : Temporal data handling using ER-Mapper 
Algorithm file. 
4. TEMPORAL DATA APPROACH FOR REMOTE 
SENSING APPLICATION 
The objective of this approach is how to support Remote 
Sensing Software with a data management tool to record image 
file location and image file information, including temporal 
informations to perform historical data query. File location is a 
file search path of the image file. In local stand-alone storage, 
this search path may contain a volume code in a directory path; 
in a network, the file search path may contain the machine name 
or an IP number or generally the URL. Using an URL to locate