|
|
Thread Tools | Display Modes |
#1
|
|||
|
|||
Rendering FITS Data
This is more of a general question about rendering data to an
image, than about FITS data per se. For astronomers, what are some of the more useful data-to-color transformations? I'm having difficulty resolving domain-specific vocabulary, so I hope I'm not being confusing -- it has certainly stymied my attempts to google this information. For instance: If you have a 2d array of 16-bit signed integer values, you can make a simple linear transformation to and 8-bit grayscale palette. I imagine log and poly transforms are of use as well. Basically, I'm looking for a discussion of useful techniques, either here, or some existing web resources. Thanks for your time and consideration. |
#2
|
|||
|
|||
[fitsbits] Rendering FITS Data
On Mon 2003-10-13T21:59:24 -0500, Grinder hath writ:
If you have a 2d array of 16-bit signed integer values, you can make a simple linear transformation to and 8-bit grayscale palette. I imagine log and poly transforms are of use as well. In the absence of the experts (see below) I suppose that the best (only?) way to determine this is to survey the existing FITS display programs. I have encountered programs that map the image to display using sqrt, square, log, linear, and various histogram equalization and/or cutoff methods. Basically, I'm looking for a discussion of useful techniques, either here, or some existing web resources. Thanks for your time and consideration. This parallels a question that has been bouncing around the FITS community -- should there be various levels of FITS conformance? In this case perhaps what would help most is a class of FITS documents which could have a status akin to the Internet (IETF) RFCs that are known as Best Current Practice (BCP). The problem with this notion is that writing and approving any such documents are mostly thankless tasks which subtract from the valuable time of the few who are interested in implementing them. If these sorts of techniques are published it is likely that many of their authors are currently at the ADASS conference, so this discussion is likely to take many days to finish. -- Steve Allen UCO/Lick Observatory Santa Cruz, CA 95064 Voice: +1 831 459 3046 http://www.ucolick.org/~sla PGP: 1024/E46978C5 F6 78 D1 10 62 94 8F 2E 49 89 0E FE 26 B4 14 93 |
#3
|
|||
|
|||
[fitsbits] Rendering FITS Data
"Steve Allen" wrote in message
rao.edu... Thanks for your response. I can't say I'm a dedicated fan, but enjoyed your appearance on "Homicide: Life on the Streets" a few years ago. On Mon 2003-10-13T21:59:24 -0500, Grinder hath writ: If you have a 2d array of 16-bit signed integer values, you can make a simple linear transformation to and 8-bit grayscale palette. I imagine log and poly transforms are of use as well. In the absence of the experts (see below) I suppose that the best (only?) way to determine this is to survey the existing FITS display programs. I have encountered programs that map the image to display using sqrt, square, log, linear, and various histogram equalization and/or cutoff methods. I've taken a look at ds9 and FITSView. My rendering methods match the output from ds9, but FITSView must mean something different by "linear" than ds9, as it's output is radically different. For example: Given a 16-bit integer data set, ds9 apparently maps the MSB of each pixel to a linear 8-bit grayscale. So values like, 0x0000 to 0x0400 appear essentially the same -- black. FITSView, though, has much better contrast for these low values even when set to linear. Examining a screenshot of the rendering shows that it is not strictly using grayscale colors. Small variations in the RGB components are present, providing a larger palette, but still appearing as a grayscale. Any idea what they're doing? Basically, I'm looking for a discussion of useful techniques, either here, or some existing web resources. Thanks for your time and consideration. This parallels a question that has been bouncing around the FITS community -- should there be various levels of FITS conformance? In this case perhaps what would help most is a class of FITS documents which could have a status akin to the Internet (IETF) RFCs that are known as Best Current Practice (BCP). The problem with this notion is that writing and approving any such documents are mostly thankless tasks which subtract from the valuable time of the few who are interested in implementing them. If these sorts of techniques are published it is likely that many of their authors are currently at the ADASS conference, so this discussion is likely to take many days to finish. That's good background. Does the FITS community have any collecting points on the web? |
#4
|
|||
|
|||
[fitsbits] Rendering FITS Data
Sorry to so egregiously pump you for links/ideas, but do you
have any suggestions as to where I can find a wide variety of test images. Google searches have turned up some goods, but nothing as rigorous as: http://www.schaik.com/pngsuite/pngsuite.html Thanks again for your consideration already rendered. |
#5
|
|||
|
|||
[fitsbits] Rendering FITS Data
On Sat, 18 Oct 2003 01:34:21 -0500, "Grinder" said:
Sorry to so egregiously pump you for links/ideas, but do you have any suggestions as to where I can find a wide variety of test images. I'd try http://www.cv.nrao.edu/fits/ for starters, and look at Don Wells' collection of FITS data: http://www.cv.nrao.edu/fits/data/. - Pat -- Patrick P. Murphy, Ph.D. Division Head, CV Computing, NRAO Home: http://goof.com/~pmurphy/ Work: http://www.nrao.edu/~pmurphy/ "Laws of nature are...just parochial by-laws in our cosmic patch" - Martin Rees |
#6
|
|||
|
|||
[fitsbits] Rendering FITS Data
Grinder wrote:
For astronomers, what are some of the more useful data-to-color transformations? I'm having difficulty resolving domain-specific vocabulary, so I hope I'm not being confusing -- it has certainly stymied my attempts to google this information. As one of those who just got back from ADASS in Strasbourg, I'll add the ways which the venerable SAOimage program (which I maintain at http://tdc-www.harvard.edu/software/saoimage/ ) converts from FITS input data to 8-bit display pixels. It supports linear (a linear scale from the minimum value at 0 and the maximum at 256), square root, log, wrap (which produces a sort of contour image), and histogram. The histogram can be clipped at the ends to emphasize the middle values. All of the code is in C in the sclmap.c file of the source code which is available through the page cited above. -Doug Mink Telescope Data Center Smithsonian Astrophysical Observatory Cambridge, Massachusetts USA |
#7
|
|||
|
|||
[fitsbits] Rendering FITS Data
"Doug Mink" wrote in message rao.edu... Grinder wrote: For astronomers, what are some of the more useful data-to-color transformations? I'm having difficulty resolving domain-specific vocabulary, so I hope I'm not being confusing -- it has certainly stymied my attempts to google this information. As one of those who just got back from ADASS in Strasbourg, I'll add the ways which the venerable SAOimage program (which I maintain at http://tdc-www.harvard.edu/software/saoimage/ ) converts from FITS input data to 8-bit display pixels. It supports linear (a linear scale from the minimum value at 0 and the maximum at 256), square root, log, wrap (which produces a sort of contour image), and histogram. The histogram can be clipped at the ends to emphasize the middle values. All of the code is in C in the sclmap.c file of the source code which is available through the page cited above. Thank you Steve, Doug and Patrick. |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
French's Primordial Study and Schramm & Turner, 1997 | greywolf42 | Astronomy Misc | 19 | July 11th 04 06:43 PM |
FAQ-2-B: sci.space.tech reading list | dave schneider | Technology | 11 | June 10th 04 03:54 AM |
FAQ-2-B: sci.space.tech reading list | dave schneider | Technology | 23 | January 20th 04 11:42 PM |
Boeing Awarded $9.2 Million to Process Radar Data from Space Shuttle Endeavour | Ron Baalke | Science | 0 | October 8th 03 05:39 PM |
Boeing Awarded $9.2 Million to Process Radar Data from Space Shuttle Endeavour | Jacques van Oene | Space Shuttle | 0 | October 8th 03 11:53 AM |