Page 173 - Digital Analysis of Remotely Sensed Imagery

138 Cha pte r F o u r

user interface, user-developed modules can be easily incorporated
into the system for better interaction with map display and access to
local and networked databases.
As a public domain system, GRASS has a limited user support
and documentation. A programmer’s manual in PDF and HTML for-
mats is publicly available for download. There is a frequently asked
questions (FAQ) document in Wikipedia that answers common ques-
tions the user may have about running GRASS programs. It is rela-
tively easy to navigate through this document to find the information
on how to perform a specific image processing task. However, more
efforts are needed to compile a comprehensive user manual. There
are certain tutorial materials in existence, but not all of them are writ-
ten in English. In these tutorials examples are provided to illustrate
how to perform certain image analysis tasks, such as image overlay/
clipping, vector map import/export, and network analysis. Nevertheless,
not all available programs/scripts are fully documented.
The limited range of GRASS image processing capability has been
improved with the addition of new programs for image enhancement,
data fusion, spectral unmixing, and georeferencing. Other recently
created functions include LiDAR data processing and DEM extrac-
tion. However, important image classification functions, such as post-
classification processing, accuracy assessment, and change detection,
that an analyst takes for granted in other commercial systems are
absent from GRASS. For these reasons, it has not found wide applica-
tions beyond certain government agencies in the United States.

4.8 Comparison
When the digital image processing systems reviewed above were ini-
tially developed, each was targeting a particular area of application
and had its unique features. Over the years these systems have under-
gone tremendous improvements and expansion in their functionality
and ease of operation, thanks to the advances in computing technol-
ogy. Upon requests from the user community, each system has per-
fected its functionality to such an extent that unique features are disap-
pearing quickly. Now most of them have become powerful systems
offering a comprehensive set of tools, with add-on modules for special-
ized applications. Increasingly, each system is becoming more and
more versatile in its analytical functions. More analyses than ever
before can be easily accomplished in the same system. Consequently,
image analysis has become a smaller component of a sophisticated
package. In terms of the data being analyzed, they have evolved from
exclusive satellite imagery to include stereoscopic aerial photographs
and radar imagery. These data are processed for image georeferencing,
mosaicking, and elevational information extraction, thus forging a
close integration of the geometric component of image analysis with
digital photogrammetry. In terms of spectral information processing,