| plotting and large data files [message #13257] |
Thu, 29 October 1998 00:00 |
David Ritscher
Messages: 30
Registered: August 1995
|
Member |
|
|
How does one handle plotting of data sets with many more points than
one can plot on the screen or print on a laser printer?
I often work with large data sets, and am hitting up against some
limitations in IDL and PV-Wave's plotting routines. In a typical plot
I wish to plot 60,000 points per line. Obviously, this is more than a
computer screen can render. There is the plotting keyword Nsum
(corresponding to !P.Nsum), which averages several points together for
each plotted point. If the raw data is very smooth,this would be
useful. Unfortunately, I'm looking at things like ECG signals, where
there are 100 or so little peaks to be seen if I plot the raw data,
which are lost if I use the Nsum keyword.
The solution for display of data where one can't afford to lose any
detail is a bit awkward:
1. calculate the exact plotting area, in pixels, of the current
graphics window
2. for each horizontal pixel (column), figure out which of the raw
data points will map into this pixel (column)
3. take the min and max of this range of the raw data points
4. for eaching pixel column on the screen, plot both a min and a max
point
This should (and seems to) produce a result identical to plotting all
the raw points. Information is never lost with this approach
(including noise information). Steps 2 and 3 were done in C for speed.
Finally, now, my questions:
1. Is there any way to do something similar with a laser printer?
Different laser printers have different mappings of pixels, sometimes
even different mappings in the two directions (for example, higher
resolution in the long axis of the paper). How would I perform the
above steps 1. and 2. with a laser printer?
This would be very useful, since I like to print one half hour of a
patients data per page. Although I'm only plotting one channel, this
still results in PostScript file sizes of about 10 MB. This tends to
choke up resources.
2. Would other users find this capability a useful thing to add to IDL
and PV-Wave basic functionality? As an alternative to the Nsum
keyword, the keyword control I would want would be to have the raw
data automatically downsamples the correct amount to correspond to the
actual pixels, and with a capability of controlling how this
downsampling is done:
* averaging
* sampling (i.e., take first sample of each of the raw data sequences)
* min and max (plot a high and low for each column, as above)
* user-defined function
David Ritscher
--
Cardiac Rhythm Management Laboratory
Department of Medicine
University of Alabama at Birmingham
B168 Volker Hall - 1670 University Boulevard
Birmingham AL 35294-0019
Tel: (205) 975-2122 Fax: (205) 975-4720
Email: david.ritscher@bigfoot.com
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
