Feature Detection in Smoothed Images
Mike Fitzpatrick wrote on Jul 21, 2013
[Moderator's Note: Posted on user's behalf]
My purpose is to smooth an optical image and check whether there is a source at a ra&dec which is got from other energy band.The source is near a galaxy's centre.
I checked web resources such as http://stsdas.stsci.edu/cgi-bin/gethelp.cgi?gauss and http://en.wikipedia.org/wiki/Gaussian_blur.
Could I ask several more silly questions?
Qa
I should smooth the image and then subtract the smoothed image from the original image ,right?
Please take a look at the png attachment named feature.png. After subtraction ,I really find some kind of feature.How to qualify the significance of the detection?
Qb
How to know the feature is a point or extended source since the feature is faint?I should make a profile by,say, imexamine ?
Qc
About the definition of convolution here,please take a look at the attachment convolution.png.We suppose a circular profile.
It means for pixel a,its new value will be got from its neighbours and their values is weighted by a Gaussian function,right?
The largest weighted value is from pixel a itself.The r value in Gaussian function is pixel distance to pixel a?
How do we get the sigma value?
Qd
At last I have to do aperture photometry.It has nothing to do with the smoothed image,right?The use of the smoothed image is just for the visualization of the feature.
So I just choose appropriate source and background apertures in the original image, and do subtraction just like the source is not in a bright background ,right?
Thank you very much.
[img:1368f64d54]http://iraf.noao.edu/irafnet_images/hang_feature.png[/img:1368f64d54]
[img:1368f64d54]http://iraf.noao.edu/irafnet_images/hang_conv.png[/img:1368f64d54]
My purpose is to smooth an optical image and check whether there is a source at a ra&dec which is got from other energy band.The source is near a galaxy's centre.
I checked web resources such as http://stsdas.stsci.edu/cgi-bin/gethelp.cgi?gauss and http://en.wikipedia.org/wiki/Gaussian_blur.
Could I ask several more silly questions?
Qa
I should smooth the image and then subtract the smoothed image from the original image ,right?
Please take a look at the png attachment named feature.png. After subtraction ,I really find some kind of feature.How to qualify the significance of the detection?
Qb
How to know the feature is a point or extended source since the feature is faint?I should make a profile by,say, imexamine ?
Qc
About the definition of convolution here,please take a look at the attachment convolution.png.We suppose a circular profile.
It means for pixel a,its new value will be got from its neighbours and their values is weighted by a Gaussian function,right?
The largest weighted value is from pixel a itself.The r value in Gaussian function is pixel distance to pixel a?
How do we get the sigma value?
Qd
At last I have to do aperture photometry.It has nothing to do with the smoothed image,right?The use of the smoothed image is just for the visualization of the feature.
So I just choose appropriate source and background apertures in the original image, and do subtraction just like the source is not in a bright background ,right?
Thank you very much.
[img:1368f64d54]http://iraf.noao.edu/irafnet_images/hang_feature.png[/img:1368f64d54]
[img:1368f64d54]http://iraf.noao.edu/irafnet_images/hang_conv.png[/img:1368f64d54]
Last post on Jul 21, 2013