Wang Song wang at
Tue Apr 20 15:39:21 EDT 1993

> Hello Wang,
> Judging by your comments on transferring files between GRASS and ERDAS
> or PCI, you are familiar with all of these packages.  By any chance,
> do you have a preference?  What would that be and why?
> Dan Bicket
> FlightSafety International
> 2590 North Highway 94
> St. Charles, MO  63301-0037
> bicket at
> (314) 925 8571

	It is always hard to answer a question which package is better because 
each package has its own strength and weakness in different aspects. To select 
a package or decide a preference, it is usually depending on the requirements
of your applications and specific interests.

	As the best of my knowledge, I would like to share some of my personnel
views on these packages with you (GRASS netter), any comments and suggestions
are greatly appreciated.

	***************  G R A S S  **************

	Though GRASS, ERDAS and PCI are all raster-based software, unlike ERDAS
and PCI, GRASS is a GIS oriented package which is definitely better than both 
ERDAS and PCI on the GIS modeling and analysis. Its public-domain source code
and data structure make it more accessible and user-friendly to the general

	The image processing functions in GRASS are relatively weak. The core 
of GRASS image processing functions are "i." functions. However, to complete a
image processing analysis, you may need to design a specific routine for your
own application with a combination of "i.","m.","r.","g." commands. In another
word, GRASS commands are not well organized for image processing applications,
although GRASS could do a lot of powerful processes on imagery as good as some
other professional image processing packages can do (for example, r.mapcalc..).
You have to be knowledgable on what you are doing and familiar with most of
the GRASS function models. It is hard for the GRASS starters.

	In a sense to image enhancement, GRASS is nothing short on spatial 
enhancement with r.mfilter and r.mapcalc. But spectraly, GRASS is far behind 
other IP packages. i.gray.scale works only on black white image. r.rescale with
equal interval stretch does not work well with the image. I have not seen the
nonlinear, hitogram equalized, pseudo-color contrast stretch functions yet which
will help to improve the GRASS image visual quality.

	***************  E R D A S  ***************

	ERDAS is currently still a giant in the image processing world. ERDAS 
7.5 version is most widely used for the ERDAS users. 7.5 version is a command 
line driven package. The good thing about ERDAS is that its functions cover
almost every corner of the image processing and the image file structures (LAN
or GIS) are more reasonable and efficient to manage multi-band imagery in 
comparason to either GRASS (image group) or ARC/INFO (image stack). 

	However, ERDAS is an expensive commercial package and not easy to learn.
I found its I/O functions are short except the Live-Link to ARC/INFO. ERDAS 
support the tape devices with its special drivers which you need to pay, but 
you could use a general UNIX "dd" command easily bypass the expensive drivers. 
ERDAS also support few other input and output devices (digitizer, plotter, 
scanner), which makes this software very hardware restricted. In addition, some
of its programs might be written in FORTRAN, the files readed by the programs
have to be in a very rigid form (each space is counted). If you need ERDAS
company help you out on I/O issues, you may have to deal with "dollars and 

	The new generation of ERDAS IMAGINE 8.x seems to work really well, but 
it is incomplete. IMAGINE has been designed with a very user friendly GUI 
(Graphical User Interface), Well-designed on-line help manual and a spatial
modeler toolbox which adopt an object-oriented concept to allow users uses its
built-in toolbox to develop their own application flow chart, then the spatial 
modeler will translate the user defined flow chart into ERDAS macro language
EML and run complete image processing routine for the users. But remember, the
IMAGINE is incomplete and the problem of limited I/O options have not be 
significantly improved so far. By any means, IMAGINE has been improved ERDAS 
a lot.

	*****************  P C I  *****************

	PCI is a canadian commercial image processing package. This package
is complete and user interface has been standarized and modualized. Also, the
PCI's aerial photo and radar image processing sub-packages are available with
no extra costs. PCI functions are very similar to those in ERDAS, but it is
relatively easy to learn and operate. PCI's multi-channel Image file (PCI.PIX) 
structure is also very similar to ERDAS multi-band image file (.LAN). PCI's
I/O functions give more options than ERDAS. 

	The front-end user interface and on-line help of PCI works fairly well,
but in a sense of X-window GUI (graphical user interface), it is not as good as
ERDAS IMAGINE. In addition, I have not seen the object-oriented user interface
design in PCI, which is used in the Spatial Modeler of IMAGINE.

	In terms of cost and performance, PCI is really not bad choice in
comparison to ERDAS 7.5 version. Also, PCI has more flexible import and export

	The unique real-time FLY module in PCI is really nice and it make 
the software very attractive to a lot of people. 

	If you are a GRASS user, PCI is more portable with GRASS than ERDAS.

	If I were you, I may select PCI instead of ERDAS as the image processing
package now. However, I will certainly keep my eye on the incomplete ERDAS 
IMAGINE.  Don't forget, ERDAS software and its file structure are still very 
popular and widely accepted, and Object-Oriented design may represent the 
future generations.


 #    #    ##    #    #   ####      Wang Song  (wang)
 #    #   #  #   ##   #  #    #     
 #    #  #    #  # #  #  #          Center for Advanced Spatial Technologies
 # ## #  ######  #  # #  #  ###     12 Ozark Hall
 ##  ##  #    #  #   ##  #    #     University of Arkansas
 #    #  #    #  #    #   ####      Fayetteville, AR 72701

INTERNET:  wang at      Tel: (501) 575-6159
                                   FAX: (501) 575-3846


More information about the grass-user mailing list