How to mosaic to a large image?
hymong wrote on Jan 22, 2008
Dear Sir/Madam,
I would like to mosaic small region astronomical images to a large one. But I didn't have any experience about this task. I think there should be much more information and some discussions here. Our system has six narrow band filters (three for emission line and three for continuum : Ha, [SII], [OIII]). Each time we make an observation, we get three images synchronously. And then we move to adjacent region to get further images and so on.
Presently, I would like to mosaic each band to a large one. Could somebody give me any suggestion about this task? Should I have to install MSCRED package? And how to do it easily? I think there is much more useful information in this forum. Please help me and give me some advices!! Thank you very much!!
Sincerely yours,
hymong
I would like to mosaic small region astronomical images to a large one. But I didn't have any experience about this task. I think there should be much more information and some discussions here. Our system has six narrow band filters (three for emission line and three for continuum : Ha, [SII], [OIII]). Each time we make an observation, we get three images synchronously. And then we move to adjacent region to get further images and so on.
Presently, I would like to mosaic each band to a large one. Could somebody give me any suggestion about this task? Should I have to install MSCRED package? And how to do it easily? I think there is much more useful information in this forum. Please help me and give me some advices!! Thank you very much!!
Sincerely yours,
hymong
Francisco Valdes wrote on Jan 22, 2008
Hi,
There are four steps to mosaicing exposures. 1) WCS calibrate the exposures. If this is not possible then determine the registration offsets. 2) Resample the exposures to a common sampling. 3) Match the intensity scales. 4) Combine the images.
Depending on the data and the accuracy needed some of these steps may not be needed. The combine step in IRAF does the final pasting together of frames provided there are only x and y integer offsets between them.
The most general case with WCS matching is best handled by MSCRED where the task uses are MSCCMATCH, MSCIMAGE, MSCIMATCH, MSCSTACK.
For non-WCS situations where there is mostly a shift and stack the tasks in IMMATCH are used including the *REGISTER variants and IMCOMBINE.
IMCOMBINE can be used directly for x/y integer shifts.
All the resampling tasks are based on GEOTRAN where they internally deal with the intricacies of making the transformation map.
From what you described of your instrument, I suspect that you need to look at the tasks in IMMATCH for simple registration and then use IMCOMBINE. This is because it is the same instrument with data taken close in time. You might be able to just use integer shifts where you would make an "offsets" file, possibly by just measuring pixel coordinates fo the same star in each image, and then use IMCOMBINE to mosaic them together.
I hope this helps. You need to dive in and play around with things. There is no single recipe for all data.
Yours,
Frank Valdes
There are four steps to mosaicing exposures. 1) WCS calibrate the exposures. If this is not possible then determine the registration offsets. 2) Resample the exposures to a common sampling. 3) Match the intensity scales. 4) Combine the images.
Depending on the data and the accuracy needed some of these steps may not be needed. The combine step in IRAF does the final pasting together of frames provided there are only x and y integer offsets between them.
The most general case with WCS matching is best handled by MSCRED where the task uses are MSCCMATCH, MSCIMAGE, MSCIMATCH, MSCSTACK.
For non-WCS situations where there is mostly a shift and stack the tasks in IMMATCH are used including the *REGISTER variants and IMCOMBINE.
IMCOMBINE can be used directly for x/y integer shifts.
All the resampling tasks are based on GEOTRAN where they internally deal with the intricacies of making the transformation map.
From what you described of your instrument, I suspect that you need to look at the tasks in IMMATCH for simple registration and then use IMCOMBINE. This is because it is the same instrument with data taken close in time. You might be able to just use integer shifts where you would make an "offsets" file, possibly by just measuring pixel coordinates fo the same star in each image, and then use IMCOMBINE to mosaic them together.
I hope this helps. You need to dive in and play around with things. There is no single recipe for all data.
Yours,
Frank Valdes
hymong wrote on Jan 22, 2008
Dearest Frank,
Firstly I am very sorry to reply late because I attended an annual meeting over past few days. And I would like to thank you so much sincerely. Thanks for your so detailed and instructive information on mosaicing exposures. It is really helpful to a newcomer like me to play this task. As you mentioned, there is no single recipe for all data. I have to dive in and play around with them. Although this task seems to look little in scientific aspects, I think it is a good example to be familiar with IRAF instructions. And the feeling to make it must be great. I will try my best to play and finish it. Thanks you very much again!!
Sincerely yours,
hymong
Firstly I am very sorry to reply late because I attended an annual meeting over past few days. And I would like to thank you so much sincerely. Thanks for your so detailed and instructive information on mosaicing exposures. It is really helpful to a newcomer like me to play this task. As you mentioned, there is no single recipe for all data. I have to dive in and play around with them. Although this task seems to look little in scientific aspects, I think it is a good example to be familiar with IRAF instructions. And the feeling to make it must be great. I will try my best to play and finish it. Thanks you very much again!!
Sincerely yours,
hymong
hymong wrote on Jan 22, 2008
Dear Frank,
I feel sorry to bother you on the mosaicking topic again. I have done some trials but I didn't know how to solve the problem.
As you mentioned, there are two ways to mosaicking exposures. One is without WCS calibration and the other is with. I have tried both.
1. In the method without WCS calibration, I used XREGISTER to compute the x and y shifts and fed the output to the parameter OFFSETS in IMCOMBINE. It didn't response error. However, when I open it with DS9, I cannot see mosaicked image. I found that the image size after IMCOMBINE is almost the same with original ones (It should be double in one dimension). I think I must ignore some parameter settings. Could you please help me?
2. In the method with WCS calibration, I used some other commands (CCMAP and CCSETWCS) to wcs each images. And then I used IMCOMBINE and set parameter OFFSETS to 'wcs' to mosaic two adjacent regions. It did generate mosaicked images. However, when I zoomed in to look the processed image, I found that it is very bad. The stars in the overlap region didn't align well at all. My first thought is that it is probably due to our large field of view, 1.5 deg by 2 deg. We have encountered the image distortion problem. I have given and discussed our result with some others. They also think so. One senior scientist tell me that I have to flatten the image. Because they have done mosaicking via their home-made program and their field of view isn't so large as us, he cannot predict more and suggest me to do it further. However, I didn't have any idea how to proceed using IRAF. Could you please give me further suggestions? I have uploaded some images to the following website for your further reference:
http://star.phys.ntu.edu.tw/~hymong/
And could you please suggest me some helpful/basic books or resources on related issue in astronomical image processing, I have the feeling that it is case by case and I cannot grasp the general/basic ideas in this fields. I am lost when I do the image process task.
I am very sorry to bother you again. Hope my problem and question won't waste much of your time. Thanks in advance!!
Sincerely yours,
hymong
I feel sorry to bother you on the mosaicking topic again. I have done some trials but I didn't know how to solve the problem.
As you mentioned, there are two ways to mosaicking exposures. One is without WCS calibration and the other is with. I have tried both.
1. In the method without WCS calibration, I used XREGISTER to compute the x and y shifts and fed the output to the parameter OFFSETS in IMCOMBINE. It didn't response error. However, when I open it with DS9, I cannot see mosaicked image. I found that the image size after IMCOMBINE is almost the same with original ones (It should be double in one dimension). I think I must ignore some parameter settings. Could you please help me?
2. In the method with WCS calibration, I used some other commands (CCMAP and CCSETWCS) to wcs each images. And then I used IMCOMBINE and set parameter OFFSETS to 'wcs' to mosaic two adjacent regions. It did generate mosaicked images. However, when I zoomed in to look the processed image, I found that it is very bad. The stars in the overlap region didn't align well at all. My first thought is that it is probably due to our large field of view, 1.5 deg by 2 deg. We have encountered the image distortion problem. I have given and discussed our result with some others. They also think so. One senior scientist tell me that I have to flatten the image. Because they have done mosaicking via their home-made program and their field of view isn't so large as us, he cannot predict more and suggest me to do it further. However, I didn't have any idea how to proceed using IRAF. Could you please give me further suggestions? I have uploaded some images to the following website for your further reference:
http://star.phys.ntu.edu.tw/~hymong/
And could you please suggest me some helpful/basic books or resources on related issue in astronomical image processing, I have the feeling that it is case by case and I cannot grasp the general/basic ideas in this fields. I am lost when I do the image process task.
I am very sorry to bother you again. Hope my problem and question won't waste much of your time. Thanks in advance!!
Sincerely yours,
hymong
Last post on Jan 22, 2008