Estimating PSNR in High Definition H.264/AVC Video Sequences Using Artificial Neural Networks
Martin Slanina, Václav Říčný · 2008
Abstract. The paper presents a video quality metric de-signed for the H.264/AVC codec. The metric operates di-rectly on the encoded H.264/AVC bit stream, parses the encoding parameters and processes them using an artifi-cial neural network. The network is designed to estimate peak signal-to-noise ratios of the video sequence frames, thus enabling computation of full reference objective qual-ity metric values without having the undistorted video ma-terial prior to encoding for comparison. We present the metric framework and test its performance for LDTV (low definition television) as well as HDTV (high definition television) video material.