| Re: UNsigned Integer Data [message #8050] |
Wed, 05 February 1997 00:00  |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
Marc Kippen <marc.kippen@msfc.nasa.gov> writes:
> Can anyone tell me how to convert between signed and unsigned
> short integer data formats in IDL?
There is no unsigned integer format in IDL. :-)
> e.g., how can I decode a byte array into an unsigned short
> integer? The FIX function automatically interprets the sign
> bit as a sign, rather than a data bit.
Marc, I'm going to assume that what you mean by this
is that you have unsigned 16-bit integers in a binary
or unformated data file. And you want to know how to
interpret this data properly in IDL.
Suppose your have a 256 by 256 array of these 16-bit
integers in your binary data file. You will do something
like this:
; Read the data as IDL signed integers (16-bit)
array = INTARR(256, 256)
OPENR, lun, unsignedIntDataFile, /GET_LUN
READU, lun, array
FREE_LUN, lun
; Convert the SIGNED integers to UNSIGNED values.
array = LONG(array) AND 'FFFF'x
Now you have an array of LONG integers, but they
have the correct unsigned values. There is no way to get
around the requirement for LONG integers unless your data
is always between 0 and 2^31-1 or 2147483647.
(If you have a 256 by 256 byte array [another way to
interpret your question], then you will want to make
array = INTARR(128,128) to read the data correctly.)
Cheers!
David
-----------------------------------------------------------
David Fanning, Ph.D.
Fanning Software Consulting
2642 Bradbury Court, Fort Collins, CO 80521
Phone: 970-221-0438 Fax: 970-221-4762
E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com
-----------------------------------------------------------
|
|
|
|
| Re: UNsigned Integer Data [message #8134 is a reply to message #8050] |
Thu, 06 February 1997 00:00  |
Struan Gray
Messages: 178
Registered: December 1995
|
Senior Member |
|
|
David Fanning, davidf@dfanning.com writes:
> array = LONG(array) AND 'FFFF'x
>
> Now you have an array of LONG integers, but they
> have the correct unsigned values. There is no way to get
> around the requirement for LONG integers unless your data
> is always between 0 and 2^31-1 or 2147483647
If you have unsigned, 2-byte integers that go:
0000 0000 0000 0000 : 0
1000 0000 0000 0000 : 32768
1111 1111 1111 1111 : 65535
All you really need to do to get signed integers is
twiddle the top bit:
1000 0000 0000 0000 : 2s-comp -32768
0000 0000 0000 0000 : 2s-comp 0
0111 1111 1111 1111 : 2s-comp 32767
Any routines that calculate a real-world value (ie to label
and axis on a graph) need to have an offset part that accounts
for the fact that 'zero' has moved, but for a lot of datasets
that's a more convenient way of handling things anyway.
Thus you can avoid the memory and time penalty of converting
to longs with something like this:
array = temporary(array) xor fix(-32768)
It definitely works faster on our macs, and since some of
our data files are pushing the memory limits anyway it saves
a lot of disk-swapping for the big ones.
Struan
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
