Quality assessment for MPEG-2 video streams using a neural network model

Quality assessment plays a crucial role in video processing and application. A no-reference objective assessment method is proposed for MPEG-2 video streams in this paper. Our model is based on an artificial neural network (ANN) with BP algorithm. At first, we directly extract various parameters from video streams and select five representative features as the inputs of our ANN model. These picked-out video features can reflect compression degree and the videos´ spatial and temporal characteristics. Then the ANN model builds the mapping relationship between these features and subjective perceived quality. The experimental results show that our model can achieve good performance for video quality prediction. A remarkable advantage of our model is bit-stream-based, which eliminate most of decoding process. It is feasible to measure the quality of many video streams/channels in parallel by our model.