Evaluation of PSNR

Lin Zhang, Dept. Computing, The Hong Kong Polytechnic University


Introduction

PSNR (Peak Singal-to-Noise Ratio) index is a traditional IQA metric.


Source Code

We used the PSNR implementation provided by Dr. Zhou Wang, which can be downloaded here https://ece.uwaterloo.ca/~z70wang/research/iwssim/psnr_mse.m.


Usage Notes

1. This implementation can only deal with gray-scale images. So, you need to convert the RGB image to the grayscale version, which can be accomplished by rgb2gray in Matlab.


Evaluation Results

The results (in Matlab .mat format) are provided here. Each result file contains a n by 2 matrix, where n denotes the number of distorted images in the database. The first column is the PSNR values, and the second column is the mos/dmos values provided by the database. For example, you can use the following matlab code to calculate the SROCC and KROCC values for PSNR values obtained on the TID2008 database:

%%%%%%%%%%%%%%%

matData = load('PSNROnTID.mat');
PSNROnTID= matData.PSNROnTID;
PSNR_TID_SROCC = corr(PSNROnTID(:,1), PSNROnTID(:,2), 'type', 'spearman');
PSNR_TID_KROCC = corr(PSNROnTID(:,1), PSNROnTID(:,2), 'type', 'kendall');

%%%%%%%%%%%%%%%

The source codes to calculate the PLCC and RMSE are also provided for each database. This needs a nonlinear regression procedure which is dependant on the initialization of the parameters. We try to adjust the parameters to get a high PLCC value. For different databases, the parameter initialization may be different. The nonlinear fitting function is of the form as described in [1].

Evaluation results of PSNR on seven databases are given below. Besides, for each evaluation metric, we present its weighted-average value over all the testing datasets; and the weight for each database is set as the number of distorted images in that dataset.

Database

Results

Nonlinear fitting code SROCC KROCC PLCC RMSE

TID2008

PSNROnTID

NonlinearFittingTID 0.5531 0.4027 0.5734 1.0994

CSIQ

PSNROnCSIQ

NonlinearFittingCSIQ 0.8058 0.6084 0.8000 0.1575

LIVE

PSNROnLIVE

NonlinearFittingLIVE 0.8756 0.6865 0.8723 13.3597

IVC

PSNROnIVC

NonlinearFittingIVC 0.6884 0.5218 0.7196 0.8460

Toyama-MICT

PSNROnMICT

NonlinearFittingMICT

0.6132

0.4443

0.6429

0.9585

A57

PSNROnA57

NonlinearFittingA57

0.6189

0.4309

0.7073

0.1737

WIQ

PSNROnWIQ

NonlinearFittingWIQ

0.6257

0.4626

0.7939

14.1381

Weighted-Average

กก

กก 0.6874 0.5161 0.7020 กก

Reference                

[1] H.R. Sheikh, M.F. Sabir, and A.C. Bovik, "A statistical evaluation of recent full reference image quality assessment algorithms", IEEE Trans. on Image Processing, vol. 15, no. 11, pp. 3440-3451, 2006.


Created on: May. 08, 2011

Last update: Aug. 04, 2011