| coregistering images - help needed [message #93775] |
Fri, 14 October 2016 00:38  |
gunvicsin11
Messages: 93
Registered: November 2012
|
Member |
|
|
I have two images I have to align one image with respect to the other,
I have two solar images size is 2048 by 2048. But the problem is in one image the data resolution is 1.0648 arcsec/pixel and other is 1.2 arcsec/pixel
So due to this in xaxis the data coverage is 1700pixels(out of 2048 pixels) in imagea
in imageb the data coverage is 1900 pixels(out of 2048 pixels)
I want imageb with resolution and data coverage as imagea.
please do help me out in this regard.
Note: I have 500 images (like imageb) which i have to bring to imagea resolution.
thanks
|
|
|
|
| Re: coregistering images - help needed [message #93778 is a reply to message #93775] |
Fri, 14 October 2016 02:27  |
Helder Marchetto
Messages: 520
Registered: November 2011
|
Senior Member |
|
|
On Friday, October 14, 2016 at 9:38:34 AM UTC+2, sid wrote:
> I have two images I have to align one image with respect to the other,
>
> I have two solar images size is 2048 by 2048. But the problem is in one image the data resolution is 1.0648 arcsec/pixel and other is 1.2 arcsec/pixel
>
> So due to this in xaxis the data coverage is 1700pixels(out of 2048 pixels) in imagea
>
> in imageb the data coverage is 1900 pixels(out of 2048 pixels)
>
> I want imageb with resolution and data coverage as imagea.
>
> please do help me out in this regard.
>
> Note: I have 500 images (like imageb) which i have to bring to imagea resolution.
>
> thanks
If the objective is to register the two images, then I would:
1) cut the images to the relevant part (imagea to 1700x1700 pixels^2 and similarly for imageb)
2) use congrid() to fix the sizes of the two images. Make sure you use a good interpolation option
3) register the images
Notice that 1700/1900 is not the same as 1.0648/1.2. In other words, the cutting the images as mentioned, will result in different data resolutions.
Or, by enlarging imagea from 1700 to 1900 pixel, you will end up with a resolution of 1.0648*1900/1700=1.19 and not 1.2.
Cheers,
Helder
|
|
|
|
| Re: coregistering images - help needed [message #93779 is a reply to message #93778] |
Fri, 14 October 2016 02:46 |
Markus Schmassmann
Messages: 129
Registered: April 2016
|
Senior Member |
|
|
On 10/14/2016 11:27 AM, Helder wrote:
> On Friday, October 14, 2016 at 9:38:34 AM UTC+2, sid wrote:
>> I have two images I have to align one image with respect to the other,
>>
>> I have two solar images size is 2048 by 2048. But the problem is in one image the data resolution is 1.0648 arcsec/pixel and other is 1.2 arcsec/pixel
>>
>> So due to this in xaxis the data coverage is 1700pixels(out of 2048 pixels) in imagea
>>
>> in imageb the data coverage is 1900 pixels(out of 2048 pixels)
>>
>> I want imageb with resolution and data coverage as imagea.
>>
>> please do help me out in this regard.
>>
>> Note: I have 500 images (like imageb) which i have to bring to imagea resolution.
> If the objective is to register the two images, then I would:
> 1) cut the images to the relevant part (imagea to 1700x1700 pixels^2 and similarly for imageb)
> 2) use congrid() to fix the sizes of the two images. Make sure you use a good interpolation option
> 3) register the images
>
> Notice that 1700/1900 is not the same as 1.0648/1.2. In other words, the cutting the images as mentioned, will result in different data resolutions.
> Or, by enlarging imagea from 1700 to 1900 pixel, you will end up with a resolution of 1.0648*1900/1700=1.19 and not 1.2.
if the metadata of your images contain the position of the solar center
and the angle indicating where solar north is, use this data instead of
trying to figure out yourselve.
You might get slightly better results, if you use a single call to
interpolate instead of using congrid and later rotate your image and
shift it fraction of a pixel.
|
|
|
|
| Re: coregistering images - help needed [message #93780 is a reply to message #93779] |
Fri, 14 October 2016 04:12 |
gunvicsin11
Messages: 93
Registered: November 2012
|
Member |
|
|
On Friday, October 14, 2016 at 3:16:05 PM UTC+5:30, Markus Schmassmann wrote:
> On 10/14/2016 11:27 AM, Helder wrote:
>> On Friday, October 14, 2016 at 9:38:34 AM UTC+2, sid wrote:
>>> I have two images I have to align one image with respect to the other,
>>>
>>> I have two solar images size is 2048 by 2048. But the problem is in one image the data resolution is 1.0648 arcsec/pixel and other is 1.2 arcsec/pixel
>>>
>>> So due to this in xaxis the data coverage is 1700pixels(out of 2048 pixels) in imagea
>>>
>>> in imageb the data coverage is 1900 pixels(out of 2048 pixels)
>>>
>>> I want imageb with resolution and data coverage as imagea.
>>>
>>> please do help me out in this regard.
>>>
>>> Note: I have 500 images (like imageb) which i have to bring to imagea resolution.
>> If the objective is to register the two images, then I would:
>> 1) cut the images to the relevant part (imagea to 1700x1700 pixels^2 and similarly for imageb)
>> 2) use congrid() to fix the sizes of the two images. Make sure you use a good interpolation option
>> 3) register the images
>>
>> Notice that 1700/1900 is not the same as 1.0648/1.2. In other words, the cutting the images as mentioned, will result in different data resolutions.
>> Or, by enlarging imagea from 1700 to 1900 pixel, you will end up with a resolution of 1.0648*1900/1700=1.19 and not 1.2.
> if the metadata of your images contain the position of the solar center
> and the angle indicating where solar north is, use this data instead of
> trying to figure out yourselve.
>
> You might get slightly better results, if you use a single call to
> interpolate instead of using congrid and later rotate your image and
> shift it fraction of a pixel.
Thanks a lot for the replies. It was really very helpful.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
