A method of estimating coding PSNR using quantized DCT coefficients

A new method of estimating coding peak signal-to-noise ratio (PSNR) without the use of reference signals is presented. Although PSNR is commonly used as a measure of the picture degradation of digitally coded video, the calculation requires source signals as a reference.Therefore, the usage of PSNR is restricted to particular applications or systems. The proposed method enables PSNR estimation based on the probability density functions of quantized discrete cosine transform (DCT) coefficients extracted from an MPEG-2 bit stream. We experimented with MPEG-2 video coding bit streams under varying quantization scheme and evaluate a new method with comparing estimated PSNRs with actual PSNRs. Experimental results indicate that the determination coefficients are higher than 0.9 This method can apply to both SDTV and HDTV, and can evaluate PSNR of every frame coded by different picture types.

[1]  Jerry D. Gibson,et al.  Distributions of the Two-Dimensional DCT Coefficients for Images , 1983, IEEE Trans. Commun..

[2]  Zhou Wang,et al.  Blind measurement of blocking artifacts in images , 2000, Proceedings 2000 International Conference on Image Processing (Cat. No.00CH37101).

[3]  Brian D. Jeffs,et al.  Adaptive image restoration using a generalized Gaussian model for unknown noise , 1995, IEEE Trans. Image Process..

[4]  Eisuke Nakasu,et al.  A Statistical Analysis of MPEG-2 Picture Quality for Television Broadcasting , 1996 .

[5]  Yong Yan,et al.  A novel distortion-quantization model and its application in low bitrate video communications , 2002, Proceedings. International Conference on Image Processing.

[6]  Fabio Bellifemine,et al.  Statistical analysis of the 2D-DCT coefficients of the differential signal for images , 1992, Signal Process. Image Commun..

[7]  Rui J. P. de Figueiredo,et al.  Maximum likelihood estimation of a class of non-Gaussian densities with application to Ip deconvolution , 1989, IEEE Trans. Acoust. Speech Signal Process..

[8]  Rui J. P. de Figueiredo,et al.  Maximum likelihood estimation of a class of non-Gaussian densities with application to deconvolution , 1987, ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[9]  Robert L. Stevenson,et al.  DCT quantization noise in compressed images , 2001, IEEE Transactions on Circuits and Systems for Video Technology.

[10]  H.R. Wu,et al.  A generalized block-edge impairment metric for video coding , 1997, IEEE Signal Processing Letters.

[11]  Gopal Lakhani,et al.  Distribution-based restoration of DCT coefficients , 2000, IEEE Trans. Circuits Syst. Video Technol..

[12]  Olivier Rioul,et al.  Image Coding with an Norm and Confidence Interval Criteria , 1998 .

[13]  Deepak S. Turaga,et al.  No reference PSNR estimation for compressed pictures , 2002, Proceedings. International Conference on Image Processing.

[14]  Xiaolin Wu,et al.  Lossless compression of continuous-tone images via context selection, quantization, and modeling , 1997, IEEE Trans. Image Process..

[15]  Didier J. Le Gall,et al.  The MPEG video compression standard , 1991, Compcon.

[16]  Ichiro Matsuda,et al.  Lossless coding using predictors and VLCs optimized for each image , 2003, Visual Communications and Image Processing.

[17]  Joseph W. Goodman,et al.  A mathematical analysis of the DCT coefficient distributions for images , 2000, IEEE Trans. Image Process..

[18]  Olivier Rioul,et al.  Image coding with an Linfinity norm and confidence interval criteria , 1998, IEEE Trans. Image Process..

[19]  Hocine Cherifi,et al.  On the distribution of the DCT coefficients , 1994, Proceedings of ICASSP '94. IEEE International Conference on Acoustics, Speech and Signal Processing.