Ask Your Question

I don't understand this values found in my results.

asked 2015-04-14 09:18:09 -0500

Diego Moreira gravatar image

updated 2015-09-30 10:42:05 -0500

I processed two images for comparate and for obtained the PSNR and MSE values for generate results for my work of similarity between two images, but, i don't understand the generates results.

When i processing my images, i found the following results:

Image 1 PSNR 11.87 MSE 4223.802679

Image 2 PSNR 26.02 MSE 162.57

Image 3 PSNR 29.95 MSE 65.72

I need to understand the results. I don't understand the results. Thank you.

edit retag flag offensive close merge delete

2 answers

Sort by ยป oldest newest most voted

answered 2015-04-15 00:41:28 -0500

tomnjerry gravatar image

PSNR is Peak signal to noise ratio and is generally used to measure the quality of reconstruction of lossy compression codecs (e.g., for image compression). The signal in this case is the original data, and the noise is the error introduced by compression.

image description

where I(i,j) is the pixel value of uncompressed image and K(i,j) is the pixel value of image after noise has been introduced. The smaller the value of MSE, the better the compression method is.

image description

where MAX is the maximum pixel value possible. If the image is represented using 8 bits, the maximum value is 255.

PSNR is an approximation to human perception of reconstruction quality.Clearly, lesser the MSE value, better the PSNR value will be. In other words, higher the PSNR Value, the better the compression technique is.

In your case, image 3 is the image most nearer to the original image as PSNR value is high and the MSE value is smallest among all others.

edit flag offensive delete link more


How to know what the best value found? How to know if the found values determine that the image has less noise and determining that they are similar? There is a range of minimum and maximum values for these metrics? Thank you.

Diego Moreira gravatar imageDiego Moreira ( 2015-04-15 10:59:03 -0500 )edit

Could you mention, how many bits are used to represent your image? The acceptable ranges for PSNR and MSE depend on your application. Generally, If it is 8-bit, then acceptable range of PSNR is between 30dB to 50dB. You can see that when the original image I (i,j) is identical to compressed image K(i,j) (theoretically), the value for MSE is 0 and thus PNSR becomes infinite.

tomnjerry gravatar imagetomnjerry ( 2015-04-16 03:45:34 -0500 )edit

My image resolution is 8 bits. tomnjerry, do you think the values I found to PSNR and MSE are acceptable?

Diego Moreira gravatar imageDiego Moreira ( 2015-04-16 07:56:17 -0500 )edit

answered 2015-04-14 09:39:28 -0500

thdrksdfthmn gravatar image

updated 2015-04-14 09:44:04 -0500

For image 1 you have a PSNR of 11.87 and a MSE of 4223.802679. This means that the Image 1 is different from the one that you compare with (but this is linked to the MSE and noise), and there is a little noise there.

For the Image 3 case, you have less noise, and the images are much similar.

edit flag offensive delete link more

Question Tools

1 follower


Asked: 2015-04-14 09:18:09 -0500

Seen: 791 times

Last updated: Apr 15 '15