Wednesday, December 03, 2008

PSNR for subjective quality measurement

Probably the most widely used objective measure for image/video quality is peak signal to noise ratio (PSNR), calculated using following equation:
 _LARGE_!_text{PSNR_{dB}} = 10_text{log}_{10}_frac{(2^n-1)^2}{_text{MSE}}
PSNR is measured on a logarithmic scale and is based on the mean squared error (MSE) between an original and an impaired image or video frame, related to (2^n-1)^2 (the square of the highest possible signal value in the image)

It is widely used as a method of comparing the 'quality' of compressed and decompressed video image. The progressively poorer image quality is reflected by a corresponding drop in PSNR. High PSNR indicated relatively high quality and low PSNR indicates relatively low quality (see below limitation)

However, PSNR measure suffers from a number of limitations. PSNR requires an 'unimpaired' original image for comparison: this may not be available in every case and not easy to verify an original image has perfect fidelity. A particular value of PSNR does not necessarily equate to an 'absolute' subject quality, for example, two images: (a)is blur in background with PSNR of 32.7dB, (b) is blur in foreground human's face with a higher PSNR of 37.5dB. PSNR measure simply counts the mean squared pixel errors and (b) is ranked a better. [ref book:video codec design]


No comments:

Post a Comment