why I am getting Inf as result of PSNR?

I used PSNR to compare between the extracted watermark(wex.jpg) and the original watermark (msg.jpg) the result is Inf could any one help me to know why and fix the result please ? the images are uploaded I used default psnr

Answer 1

[Image Analyst]

Since the definition is

PSNR=10log10(peakval^2/MSE)

and the MSE is zero if the two images are identical, then why does that surprise you? YOu know that one over zero is infinity, right? And that log of infinity is also infinity? What do you think it should be?

[Eliza]

yes sure, but I surprised because I did not think that the extracted watermark will be exactly identical to the original WM. Also I ran the code before and it was giving me some values for psnr =17.366 the after some modifications psnr=Inf. Is there any way to measure the percentage of identical between the images? I wanna a value gives me that there are Identical.

[Eliza]

thanks sir I found your answer here. https://uk.mathworks.com/matlabcentral/answers/266446-how-to-calculate-degree-of-similarity-between-two-images

[Manar]

Hi Eliza,

Did you find how to represent (Inf) as value?

I need it for my research purpose.

Thank you

[Walter Roberson]

percentage_identical = mean2(Image1 == Image2) * 100

However, you have to think about whether this is what you want if you are comparing color images. If you have two images, and the red and green panes are identical between the two images, but the blue plane differs for some locations, then are you okay with "red identical" and "blue identical" count as +2 out of a possible +3 available for that pixel? Or do you want to say that the pixel should not be considered at all identical if any of the color panes are different?

percentage_identical = mean2(all(Image1 == Image2,3)) * 100