PSNR

This function displays the PSNR (peak signal-to-noise ratio) between two images.

Description

This function displays the PSNR (peak signal-to-noise ratio) between two images. The answer is in decibels (dB).

PSNR is very common in image processing. A sample use is in the comparison between an original image and a coded/decoded image. Typical quoted PSNR figures are in the range +25 to +35dB.

The syntax for this file is PSNR(A,B), where A and B are MATLAB Intensity Images, with matrix-elements in the interval [0,1]

The function can be adapted for images with 256 gray levels.

• Acknowledgements: This submission has inspired the following:; mmvn_toolkit
• MATLAB release: MATLAB 7.2 (R2006a)

[Shaikh Hussain]

 

[febrian taufiq]

 

[taufiq FEBRIAN]

 

[Joel Schaere]

This is Ok, not perfect, but suitable for use

[susette dayanghirang]

 

[Wei Ting]

3 PSNR values are shown.
Are they the PSNR of YUV or RGB?

[Din bhari]

I highly Apprciate it

[VEERAMALLY RAVIKUMAR]

 

[jonathan haffarei]

very simple and effective

[Michael Kleder]

Nice. Well commented.

[VENKATA KARTHIK PRABALA]

easy to understand and implement. good idea to can this.

[Jeny Rajan]

Simple and fast

[Feng Yang]

great

[bassem sheta]

what if the two images are not between [0,1]

[jameer iqbal]

 

[sankar kumar]

its nice

[ichi maeda]

This PSNR is not accurate!!
Peak is always fixed at 1.

[Ramraj Pandey]

good... we calculate PSNR with signals normalized.

[weiss Pierre]

Very poor implementation. It just works for images, peak is fixed to one. Rather than max(max(A)), the author should prefer max(A(:)) which works for images in any dimension.

[weiss Pierre]

I don't have time to do something better, but rather use that code, which may not be perfect, but which is already more general :

%function [SNR,MSE]=PSNR(u0,u)
function [SNR,MSE]=PSNR(u0,u)

[ny,nx]=size(u0);
A=max(max(u0(:)),max(u(:)));
dif=u0(:)-u(:);

MSE=mean(dif.^2);

SNR=10*log(A*A/MSE)/log(10);

[justin lee]

good thank u

[prabhat srivastava]

 

[liu qiang]

 

[burepalli venkata srinivasarao]

i want the psnr mat lab code

[Tiandi Duan]

 

[Luren Jia]

img1 = im2double(imread('img1.bmp'));
img2 = im2double(imread('img2.bmp'));

PSNR = 10*log10(1/mean((img1(:)-img2(:)).^2));

compare mine coding with yours which one is correct?

[himanshu sharma]

good

[Muhammad Naveed Ahmed]

Excellent and thanks

[Aon S.]

 

[bigforg ‡]

 

[hbu ]

good