PSNR Estimation Scheme using DCT Coefficient Distribution in H.264 Decoder
Seung-Yong Shim, Ju-Heon Seo, Jong-Ki Han · ITC-CSCC :International Technical Conference on Circuits Systems, Computers and Communications · 2007
Peak Signal-to-Noise Ratio (PSNR) is commonly used as a measure for the quality of the reconstructed picture where original picture is used as reference data. Since the original picture data is not available in video decoder, the receiver including video decoder cannot evaluate the quality of the reconstructed image in the view point of PSNR. In this paper, we propose an efficient method to estimate PSNR using statistical distribution of discrete cosine transform (DCT) coefficient in H.264 decoder. Whereas the distribution of DCT coefficients was approximated with Laplacian model in the previous studies, we use Cauchy distribution model since it is more appropriate for data compressed by H.264 standard than other models. The proposed estimation process consists of two phases: a) prediction for PDF (Probability Density Function)s of AC coefficients of original video data based on the quantized DCT coefficients decoded in the receiver; b) estimation of PSNR using the predicted PDF for AC coefficients. We show experimental results which demonstrate the effectiveness of the proposed algorithm.