No reference estimation of the coding PSNR for 4K-UHD videos

With the wide-spread use of digital videos, quality considerations have become essential, and industry demand for video quality measurement standards is rising. In this paper, we proposed a no reference estimation of the coding PSNR for 4K-UHD videos in the compressed domain. As video coding parameters which can reflect the degree of distortion caused by compression can be extracted in compressed domain, we firstly extracted some feature parameters from 4k-UHD compressed videos. Then the parameters were applied to train the proposed model with the corresponding PSNR using multi-linear algorithm. After the training process, estimated PSNR can be acquired using the trained model. The experimental results show that our model can achieve a good performance for the 4K-UHD video streams with the corresponding PSNR.