how to calculate actual snr of image in matlab?
106 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
vipul utsav
el 4 de Dic. de 2012
Comentada: Alan Keenan
el 20 de Sept. de 2021
how to calculate actual snr of image in matlab?
0 comentarios
Respuesta aceptada
Image Analyst
el 4 de Dic. de 2012
Get the signal - that's your "true" noiseless image.
Get the noise - that's your actual noisy image minus the "true" noiseless image.
Divide them element by element, then take the mean over the whole image.
12 comentarios
Image Analyst
el 5 de Dic. de 2012
Well, what if your signal is 100 everywhere, and your noisy image is 100 plus or minus 100, so it goes from 0 to 200. Don't you agree that that is a very very noisy image and it should have a low signal to noise ratio? It's total noise - all over the place. But it's mean is still 100 - same as your signal. So just looking at the means, you can't tell the difference between your perfect noise-free signal and the incredibly noisy signal. How is that any good?
By the way, you can always omit pixels where the noise is zero if you want.
Más respuestas (1)
Dhurgham Al-karawi
el 29 de Ag. de 2017
Editada: Image Analyst
el 29 de Ag. de 2017
Hi,
Here you are.
img=imread('.JPG');
img=double(img(:));
ima=max(img(:));
imi=min(img(:));
ims=std(img(:));
snr=10*log((ima-imi)./ims);
4 comentarios
Image Analyst
el 17 de Oct. de 2019
A signal to noise ratio requires both a signal, and noise. If you don't have each of those, you can't do it. However there are single image quality metrics. Try this link.
Alan Keenan
el 20 de Sept. de 2021
Perform a histogram on your image and then decide at what level you consider to be signal, i.e., above a value is signal and below that value is noise. If your image is in dB then you might use -6dB as the threshold value. In my case my max signal value is always 0dB as it has been normalised to the max value and converted to dB. So you can say anything less than -6dB is noise and anything greater than -6dB is signal. An easy way to calculate the noise floor is rms(rms(image));. This means that if your signal has a max value of 0dB then the peak SNR for your signal = 0dB + the value of the rms calculation.
Ver también
Categorías
Más información sobre Image Processing Toolbox en Help Center y File Exchange.
Productos
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!