i.pca, r.mapcalc, i.rectify

[Michael Shapiro]

(1) i.pca was written rather poorly. It  reads  the  entire  maps
into  memory.   It  didn't  need  to do this and should be fixed.
Hopefully this will be one of the "bugs" that gets fixed for 4.1

(2) maps are considered integers and  hence  band3/band1  doesn't
produce   floating   point   values.   But   even  if  you  wrote
float(band3)/band1 the result still  has  to  be  rounded  to  an
integer.  Unles  you  happend  to get ratios like 255/64, I'm not
supprised that you only get 0 and 1 (but check youar  region  and
if  you  have  a  MASK  set).   I   recommend   that   you   form
band   ratios   by (255.0*band3)/band1

(3) i.points and  i.rectify  use  a  linear  tranformation  at  a
present. The Soil Conservation Service is adding second and third
order polynomial transformations to these  tools  (rumor  has  it
they  are  adding  some  ability  to use/convert vectors files as
well).

|
|Hi (Y')all, 
|
|I have a few questions concerning some problems that I'm having using
|GRASS. Hopefully some one out there might know of solutions to the 
|following:
|
|1) I have being trying to use "i.pca" to transform 4 entire SPOT
|   images (52 MBytes each) but when I try to do so the job gets killed by
|   the system. Is this a normal response when trying to carry out PCA
|   on a large dataset or is it related to the problems with the porting of 
|   GRASS to the RISC 6000 platform?
|
|2) I have been trying to ratio two SPOT images using "r.mapcalc"
|   as follows r.mapcalc ratio=band3/band1 but the output image only
|   has minimum value=0, maximum value=1. I have checked the band
|   statistics for both bands and although I wouldn't expect the maxmimum 
|   theoretical range of 256*256 values I certaintly expected a greater
|   range of values than 1 or 0. Is there something I need to do (like set
|   number of decimal places or reference a ratio lookup table) to carry out
|   image ratioing?
|
|3) I was recently using "i.target,i.points and i.rectify" to reference a 
|   previously (State Plane) rectified SPOT image in ROW, COL format to 
|   State Plane coordinates. I would like to know if the i.rectify algorithm
|   uses a third order polynomial to rectify the input images or whether
|   it matches the most appropriate transformation equation to the inputted
|   control point coordinates?
|
|
|Thanks,
|
|Tony Gale
|Research Specialist
|City of Austin Electric Utility
|Austin, Texas
|
|
|



Michael